Neurocognitive Disturbance in OSA i
Neurocognitive Disturbance in Obstructive Sleep Apnoea: Mechanisms of Harm
Michelle Olaithe
University of Western Australia
This thesis is submitted in partial fulfilment of the requirements for the degree of
Doctor of Philosophy (Ph.D.) at the University of Western Australia
School of Psychology
The University of Western Australia
April 2014
Word count: 29,447
Declaration
“I declare that this report is an original piece of research, conducted by myself
and does not contain data or materials which have been previously submitted by
myself or anyone for academic credit. I further declare that this report does not
contain any materials previously presented by myself or another person, except
where due reference is made in the text. This study was conducted with the approval
of the Human Research Ethics Committee of the University of Western Australia.”
Name: Michelle Olaithe
Date:
Signature:
Neurocognitive Disturbance in OSA ii
Abstract
The overarching aim addressed by this thesis was the investigation of the
relationship between cognitive dysfunction and mechanisms of harm (sleep
fragmentation and hypoxia) in Obstructive Sleep Apnoea (OSA). It begins with a
general introduction to the nocturnal features and cognitive profile of OSA (Chapter
1), continues with three research studies (reported over Chapters 2, 4 and 5, Chapter 3
details methods and recruitment), and concludes with a general discussion (Chapter
6).
Study 1 (Chapter 2) considered the profile of executive dysfunction in
individuals with OSA. OSA is a frequent and often under-diagnosed condition that is
associated with upper airway collapse, oxygen desaturation, and sleep fragmentation
leading to cognitive dysfunction. There is good meta-analytic evidence that sub-
domains of attention and memory are affected by OSA. However, a thorough
investigation of the impact of OSA on different sub-domains of executive function
had yet to be conducted. Study 1 (Chapter 2) investigated the impact of untreated and
treated OSA, in adult patients, on five, theorised, sub-domains of executive function.
An extensive literature search was conducted of published and unpublished materials,
returning 35 studies that matched selection criteria. Meta-analysis was used to
synthesise the results from studies examining the impact of OSA on executive
functioning compared to controls (21 studies) and before and after treatment (19
studies); 5 studies met inclusion in both categories. All domains of executive function
(Shifting, Updating, Inhibition, Generativity and Fluid Reasoning) demonstrated
medium to very large impairments in OSA independent of age, and disease severity.
All domains improved (small to medium effects) with CPAP treatment, and this
Neurocognitive Disturbance in OSA iii
improvement was not moderated by age or disease severity. Further studies are needed
to explore the extent of primary (neural damage in a region, with corresponding
behavioural dysfunction, e.g., damage to areas responsible for memory, and
demonstrated memory difficulties) or secondary nature (neural damage resulting in
secondary dysfunction, e.g., damage in areas responsible for attention control,
impacting on memory capacity) of these deficits, and the impact of age and pre-morbid
ability (cognitive reserve).
Chapter 3 reports the recruitment procedures, participants and methods used in
this thesis.
Study 2 (Chapter 4) assessed the measurement of sleepiness using the Epworth
Sleepiness Scale (ESS) prior to inclusion of this construct as a covariate in Study 3.
The ESS is a widely used tool for measuring sleepiness. In addition to providing a
single measure of sleepiness (a one factor structure), the ESS also has the capacity to
provide additional information about specific factors that facilitate sleep-onset,
including a person’s posture, activity and environment. These features of sleepiness
are referred to as somnificity. Study 2 (Chapter 3) evaluated the fit of a 1-factor
structure (sleepiness) and a 3-factor structure (reflecting low, medium and high levels
of somnificity) for the ESS using Confirmatory Factor Analysis (CFA). Two samples
(a community sample N = 356 and a clinical sample N = 679) were administered the
ESS. In both samples, a 3-factor structure (community sample adjusted X2 = 2.95,
RMSEA = .07, CFI = .95; clinical sample adjusted X2 = 3.98, RMSEA = .07, CFI =
.98) provided a level of model fit that was at least as good as the 1-factor structure
(community sample adjusted X2 = 5.01, RMSEA = .11, CFI = .87; clinical sample
adjusted X2 = 8.87, RMSEA = .11, CFI = .92). In addition to a single measure of
Neurocognitive Disturbance in OSA iv
sleepiness, the ESS can provide subscale scores that relate to three underlying levels
of somnificity. These findings suggest that the ESS can be used to measure an
individual’s overall sleep propensity as well as more specific measures of sleep
propensity in low, moderate and high levels of situational somnificity. Due to the
findings of this chapter, the ESS was included as a covariate in Study 3, run in the
model as a single factor, and as three factors to evaluate model fit.
Study 3 (Chapters 5) determined the influence of hypoxia and sleep
fragmentation on cognition in OSA, while controlling for potentially confounding
variables including subjective sleepiness (using results from Study 2), age and
premorbid intelligence. Participants with and without OSA (N = 150) were recruited
from the general community and a tertiary hospital sleep clinic. All underwent
comprehensive, laboratory-based polysomnography and completed assessments of
cognition including attention, short-term & long-term memory and executive function.
Structural Equation Modelling (SEM) was used to construct a theoretically driven
model to examine the relationships between hypoxia and sleep fragmentation, and
cognitive function. Increased sleep disturbance was a significant predictor of
decreased attention (p = .04) and decreased executive function (p = .05), after
controlling for IQ. No significant predictors of memory function were found, and
hypoxia was not related to any cognitive domain. Controlling for age removes the
significant relationships between sleep fragmentation, attention and executive
function.
The implications of the findings are discussed in detail in Chapter 6. This chapter
elucidates the strengths, originality of this thesis, and a range of suggestions
concerning how future research can move forward from the present research program.
Neurocognitive Disturbance in OSA v
Table of Contents
Abstract ............................................................................................................... ii
Table of Contents ............................................................................................... vi
List of Tables ..................................................................................................... ix
List of Figures ..................................................................................................... x
Acknowledgements............................................................................................ xi
Statement of Contribution................................................................................. xii
Publications....................................................................................................... xv
List of Abbreviations ........................................................................................ xx
Chapter 1 General Introduction .......................................................................... 1
Adult Obstructive Sleep Apnea ....................................................................... 3
Cognition in OSA ............................................................................................ 4
Conceptual models of cognitive harm in OSA ............................................... 5
Measuring nocturnal disturbance in OSA ....................................................... 8
Measurement of cognition in OSA ............................................................... 10
Use of problematic statistical methods ......................................................... 11
Inter-individual differences moderating the impact of OSA on cognition ... 12
Summary and contents of chapters ............................................................... 13
Chapter 2 Executive Function in OSA ............................................................. 16
Preface ........................................................................................................... 16
Abstract ......................................................................................................... 17
Neurocognitive Disturbance in OSA vi
Introduction .................................................................................................. 19
Method .......................................................................................................... 22
Results ........................................................................................................... 30
Discussion .................................................................................................... 37
Chapter 3 Recruitment, Participants and Materials .......................................... 44
Introduction .................................................................................................. 44
Recruitment ................................................................................................... 45
Materials ....................................................................................................... 48
Questionnaires ........................................................................................... 48
Neurocognitive assessment ....................................................................... 52
Sleep metrics ............................................................................................. 57
Conclusions .................................................................................................. 66
Chapter 4 Examining the utility of the ESS ..................................................... 67
Preface .......................................................................................................... 67
Abstract ......................................................................................................... 69
Introduction ................................................................................................... 70
Method .......................................................................................................... 71
Results ........................................................................................................... 77
Discussion ..................................................................................................... 80
Chapter 5 Relationship between OSA severity and cognition ......................... 86
Preface .......................................................................................................... 86
Neurocognitive Disturbance in OSA vii
Abstract ......................................................................................................... 87
Introduction ................................................................................................... 88
Method .......................................................................................................... 91
Results ........................................................................................................... 96
Discussion ................................................................................................... 107
Chapter 6 General Discussion ........................................................................ 115
Overview of findings .................................................................................. 115
Discussion of findings ................................................................................. 117
Implications for clinicians ........................................................................... 124
References....................................................................................................... 128
Appendix 1...................................................................................................... 142
Appendix 2...................................................................................................... 163
Appendix 3...................................................................................................... 164
Neurocognitive Disturbance in OSA viii
List of Tables
Table 1 Subdomains of executive function and tests ....................................... 28
Table 2 Effect sizes for OSA to control and pre to post groups ....................... 33
Table 3 Demographic data for the Clinical and Community participants ........ 47
Table 4 Measurements obtained during polysomnography.............................. 59
Table 5 Sleep stages in healthy adults .............................................................. 64
Table 6 ESS total and subscale scores .............................................................. 77
Table 7 Fit indices for the one and three factor ESS models ........................... 79
Table 8 Descriptive statistics for sleep and cognitive measures ...................... 97
Table 9 Descriptive statistics for the clinical and community samples ............ 99
Table 10 Fit indices for Model 1 ad Model 2 ................................................. 106
Neurocognitive Disturbance in OSA ix
List of Figures
Figure 1 Flow chart of study inclusion and exclusion ...................................... 23
Figure 2 Proposed 1 factor structure of the ESS............................................... 75
Figure 3 Proposed 3 factor structure of the ESS............................................... 76
Figure 4 Model 1 – Accounting for estimated IQ........................................... 101
Figure 5 Model 2 – Accounting for estimated IQ and age ............................. 104
Neurocognitive Disturbance in OSA x
Acknowledgments
The work in this thesis was supported by an Australian Postgraduate Award, awarded by the
Australian Government, the University of Western Australia ‘Top-up’ Scholarship, and funds from
the School of Psychology, University of Western Australia. I gratefully acknowledge the assistance
and support of so many people I am blessed to know, far too many to mention here. I could write a
thesis of dedication to you all!
My wonderful supervisors, friends, and mentors Professor Romola Bucks, and Winthrop
Professor Peter Eastwood, you have taught me so much more than research and writing, and I will
carry these lessons and memories forever. Thank you for your patience and guidance.
My wonderful daughter, Madelynn, and the love of my life, Jesse, who are my life, and have
taught me play is far more important than work.
My beautiful supportive Mother, Ann Leaver, and Sister, Leoni Leaver, who cooked many
wonderful meals, dismissed my absence at family parties, suffered through much grumpiness and still
rocked up at my door with coffee, flowers, and encouragement.
To the beautiful people in my life who grace me with their wonderful cooking, terrible stories
and exceptional warmth: we have known each other since we were little, I hope we know each other
‘till we grow little again, Jennifer Free and Kimberley Pearman.
The amazing students, who anguished with me when I anguished, laughed when I laughed, and
helped me hunt down the crickets plaguing our room because of Colin MacLeod’s anxiety studies;
Laila Simpson, Alea Losch, Heather Liebregts, Melissa Burgess, Cindy Cabeleira, Alix Mellor, and
Ines Pandzic.
Finally, huge thanks to Professor David Hillman, Christine McGuire, and the wonderful staff
at Sir Charles Gairdner Hospital, West Australian Sleep Disorders Institute and Sleep Clinic, who
gave me so many hours and so much guidance on the right buttons to push (people and machines).
Neurocognitive Disturbance in OSA xi
Statement of Original Contribution
The work contained in this thesis has not previously been submitted for a
degree or diploma at any other higher education institution. This thesis is entirely my
own work and has been accomplished during enrolment in the degree of Doctor of
Philosophy (Psychology). All external sources have been acknowledged.
Components of the research were conducted in collaboration with other
researchers or institutions, as part of a larger study titled “Predicting Usage of
Continuous Positive Airway Pressure in Obstructive Sleep Apnoea” (PUCOSA),
directed by Professor Romola Bucks and Professor Timothy Skinner, in
collaboration with Professor David Hillman, West Australian Sleep Disorders
Research Institute, Sir Charles Gardiner Hospital, Winthrop Professor Peter
Eastwood, the Centre for Sleep Science, and the West Australian Participants Pool,
as outlined in the Acknowledgements section. However, the design of the current
project, data analysis, and the preparation of this thesis have been completed by the
candidate alone, with guidance from her supervisors.
As part of “PUCOSA”, the student carried out the following tasks: trained
three other researchers to administer assessments and to carry out study
administration; assisted with drafting the ethics committee application/amendments;
co-designed the protocol for contacting and testing participants; co-created a
standardised neuropsychological assessment manual; set up and configured the
computerised assessments used; created and administered an online survey;
contacted by phone and post numerous study participants in bi-weekly recruitment
sessions (for a total of 24 months); conducted the neuropsychological testing of 23
participants at baseline and at three months follow-up (total of 69 hours of face-to-
face participant testing); conducted 24 over-night sleep studies at the centre for sleep
Neurocognitive Disturbance in OSA xii
science (total of 288 hours of testing); conducted weekly visits to Sir Charles
Gardiner Hospital to collect questionnaire data; gave one presentation to the hospital
staff about study results/updates and to obtain feedback from the staff so as to review
and refine study procedures; gave seven public presentations to Rotary, Lions, and
adult education centres as part of community service and recruitment; entered
neuropsychological assessment data into SPSS; assisted with the creation of a data
entry handbook, defining variables and variable labels and their coding; approached
participants at the hospital to complete PUCOSA questionnaires; liaised with the
hospital about retrieval of descriptive and medical history data from hospital
databases, including assisting with extracting data; and organised, attended and/or
contributed to monthly PUCOSA management meetings for a total of 36 months.
The articles that were published as a result of the work undertaken for this
thesis are included in chapters 2, 4 and 5. The student undertook a large portion of
the data collection, completed all analyses, formulated and wrote the papers. The
other authors on the papers provided intellectual input, supervised data collection
procedures, advised on the analyses and interpretation, and assisted with formulation
and editing of the final papers.
More specifically, in chapter 2, the author RSB helped to provided intellectual
input into the conceptualisation of the concept, conducted quality assessment and
categorisation of the papers included in the final analysis, and edited the manuscript.
In chapter 4, the authors RSB and TS provided intellectual input, advised on
the analyses and interpretation, and assisted with editing the manuscript. The author
JC assisted with data collection and edited the final manuscript. The author PE
provided intellectual input and edited the final manuscript.
Neurocognitive Disturbance in OSA xiii
In chapter 5, the author RSB provided intellectual input, advised on the
analyses and interpretation, and assisted with editing the manuscript. The authors
TS, DH and PE provided intellectual input and edited the final manuscript.
In chapter 6, the author RSB provided intellectual input, and assisted with
editing the manuscript. The author PE assisted with editing the manuscript.
Neurocognitive Disturbance in OSA xiv
Publications
Components of this thesis presented at conferences
Olaithe, M., Eastwood, P. R., Skinner, T. C., Hillman, D. & Bucks, R.S., Cognitive
dysfunction in Obstructive Sleep Apnoea; Mechanisms of harm. International
Neuropsychological Society (INS) 2013 Mid-Year Meeting, Amsterdam, The
Netherlands. July 10-13, 2013. Journal of the International
Neuropsychological Society, 19, 49. doi:10.1017/S1355617713001033.
Olaithe, M., & Bucks, R.S., A Meta-analysis of executive dysfunction in OSA:
Before and after treatment, and correlates of nocturnal symptoms, Sleep up
Top, Australasian Sleep Association (ASA), Darwin, 10-13 October 2012.
Olaithe, M., Skinner, T.C., Clarke, J., Eastwood, P. & Bucks, R.S., What does the
Epworth measure? A confirmatory factor analysis of the Epworth Sleepiness
Scale (ESS). Sleep Down Under 2011 – Sleep in the City, Australasian Sleep
Association (ASA), Sydney, 27-29 October, 2011. Journal of Sleep
Research, 20(Suppl. s1), Po65.
Other abstracts published during the course of this thesis
Olaithe, M., Holt, A., Wallace, A., Kashyap, S., Hatton, J., Eastwood, P., Skinner,
T.C., Wesnes, K., & Bucks, R.S., Investigating the relationship between
memory accuracy and speed of memory retrieval in Obstructive Sleep
Apnoea (OSA). Sleep Down Under 2011 – Sleep in the City, Australasian
Sleep Association (ASA), Sydney, 27-29 October, 2011. Journal of Sleep
Research, 20(Suppl. s1), 109.
Bucks, R.S., Olaithe, M., Eastwood, P. R., Skinner, T. C., & Hillman, D. (2013)
Impact of Sleep Disordered Breathing on self-reported memory function: It’s
Neurocognitive Disturbance in OSA xv
more than just being old and sad. International Neuropsychological Society
(INS) 2013 Mid-Year Meeting, Amsterdam, The Netherlands. July 10-13,
2013. Journal of the International Neuropsychological Society, 19, 107.
doi:10.1017/S1355617713001033.
Holt, A., Olaithe, M., Wallace, A., Hatton, J., Kashyup, S., Shield, H., Skinner, T.C.,
Eastwood, P., Hillman, D., Wesnes, K., & Bucks, R.S. (2011). Intra-
individual variability in cognitive response time before and after CPAP
treatment for obstructive sleep apnoea. 17th Annual College of Clinical
Neuropsychologists Conference, Sydney, NSW, 2-5 November, 2011.
Bucks, R.S., Olaithe, M., Marshall, MM. (2011). Self-reported memory function in a
community sample: More sleep difficulties lead to lower memory ability
(Self-appraisal of one’s memory capabilities). Sleep Down Under 2011 –
Sleep in the City, Australasian Sleep Association (ASA), Sydney, 27-29
October, 2011. Journal of Sleep Research, 20(Suppl. s1), Po68.
Mellor, A., Olaithe, M., Waters, F. & Bucks, R.S. (2011). Age-related changes in
sleep and their relationship to depression. Sleep Down Under 2011 – Sleep in
the City, Australasian Sleep Association (ASA), Sydney, 27-29 October,
2011. Journal of Sleep Research, 20(Suppl. s1), Po71.
Bucks, R.S. McNeil, L., De Regt, T. Mellor, A., Phang, J., Olaithe, M., et al. (2010).
Memory complaints in obstructive sleep apnoea. Australasian Sleep
Association (ASA), Sleep Down Under 2010 - Biodiversity of Sleep.
Christchurch, New Zealand, 21-23 October 2010. Journal of Sleep and
Biological Rhythms, 8(Suppl. 1), A31.
Skinner, T.C., Phang, J., McNeil, L., De Regt, T., Mellor, A., Olaithe, M. et al.
(2010). Patients’ beliefs about their sleep problems. Australasian Sleep
Neurocognitive Disturbance in OSA xvi
Association (ASA), Sleep Down Under 2010 - Biodiversity of Sleep.
Christchurch, New Zealand, 21-23 October 2010 Journal of Sleep and
Biological Rhythms, 8(Suppl. 1), A54.
McNeil, L., Skinner, T.C., De Regt, T., Phang, J., Mellor, A., Olaithe, M., et al.
(2010). Illness perceptions in obstructive sleep apnoea population: patients
do not believe that their symptoms are due to their sleep problems.
Australasian Sleep Association (ASA), Sleep Down Under 2010 -
Biodiversity of Sleep. Christchurch, New Zealand, 21-23 October 2010,
Journal of Sleep and Biological Rhythms, 8(Suppl.1), A54.
De Regt, T.L., Skinner, T.C., Mellor, A., Holt, A., Olaithe, M., McNeil, L., et al.
(2010). Cognitive deficits in Obstructive Sleep Apnoea and their relationship
to disease severity. Cognitive deficits in obstructive sleep apnoea and their
relationship to disease severity. 16th Annual Conference of the APS College
of Clinical Neuropsychologists (CCN) Conference, Fremantle, 30 September
- 2 October 2010.
De Regt, T.L., Skinner, T.C., Mellor, A., Holt, A., Olaithe, M., McNeil, L., et al.
(2010). Cognitive deficits in Obstructive Sleep Apnoea. University of
Western Australia School of Psychology Conference, October, 2010, Perth,
Western Australia.
De Regt, T.L., Skinner, T.C., Mellor, A., Holt, A., Olaithe, M., McNeil, L., et al.
(2010). Cognitive deficits in obstructive sleep apnoea. Australasian Sleep
Association (ASA), Sleep Down Under 2010 - Biodiversity of Sleep.
Christchurch, New Zealand, 21-23 October 2010. Journal of Sleep and
Biological Rhythms, 8(Suppl. 1), A29. Xiv.
Neurocognitive Disturbance in OSA xvii
McNeil, L.D., Skinner, T.C., De Regt, T.L., Phang, J., Mellor, A., Olaithe, M.,
Eastwood, P., Whitworth, S., Holt, A., Nienaber, A., Hillman, D. & Bucks,
R.S. (2010). Illness representations in Obstructive Sleep Apnoea: patients’
symptom identities. Australasian Sleep Association (ASA), Sleep Down
Under 2010 - Biodiversity of Sleep. Christchurch, New Zealand, 21-23
October 2010, Journal of Sleep and Biological Rhythms, 8(Suppl. 1), A54.
Peer reviewed publications
(Publications from this thesis are marked with an **)
(Chapter 5) ** Olaithe, M, Skinner, T., Hillman, D., Eastwood, P. & Bucks, R.,
(Submitted Dec 2013) Cognitive Dysfunction in Obstructive Sleep Apnoea;
evaluating the contributions of sleep fragmentation and hypoxia. Sleep and
Breathing
(Chapter 2) ** Olaithe, M, & Bucks, R. (2013). Executive function in OSA: A Meta-
analysis. Sleep, 36(9), 1297-1305. http://dx.doi.org/10.5665/sleep.2950.
Bucks, R., Olaithe, M & Eastwood, P. (2013) Neurocognitive function in
Obstructive Sleep Apnea – a Meta-review. Invited review for "Translational
Advances in Respiratory Diseases”. Respirology, 18(1), 61-70. doi:
10.1111/j.1440-1843.2012.02255.x.
(Chapter 4) ** Olaithe, M., Skinner, T.C., Clarke, J., Eastwood, P. & Bucks, R.S.
(2012) What does the Epworth measure? A confirmatory factor analysis of
the Epworth Sleepiness Scale. Sleep and Breathing, 17(2), 763-9. doi:
10.1007/s11325-012-0763-6.
Skinner, T.C., McNeil, L., Olaithe, M., Eastwood, P., Hillman, D.R., Phang, J., De
Regt, T., & Bucks, R.S. (2013) Predicting uptake of Continuous Positive
Airway Pressure (CPAP) therapy in Obstructive Sleep Apnoea (OSA): a
Neurocognitive Disturbance in OSA xviii
belief-based theoretical approach, Sleep and Breathing; 17, 1229-1240. doi:
10.1007/s11325-013-0828-1.
Mellor, A., Olaithe, M., McGowan, H., Waters, F. & Bucks, R.S. (2013). Sleep and
aging: Examining the effect of depression and sleep-disordered breathing.
Behavioral Sleep Medicine. EPub ahead of print. doi: 10.1007/s11325-013-
0828-1.
(Chapter 6) **Olaithe, M., Bucks, R. & Eastwood, P.E. (Submitted 15/04/2014).
Obstructive Sleep Apnoea (OSA), affecting cognition while we’re in the
dark: Implications for practitioners. Translational issues in psychological
science – Invited review for special edition “The science of sleep”.
Neurocognitive Disturbance in OSA xix
List of Abbreviations
AHI Apnoea Hypopnoea Index
BMI Body Mass Index
CDR Cognitive Drug Research Battery
C-FIT Culture Fair Intelligence Test
COWAT Controlled Oral Word Association Test
CPAP Continuous Positive Airway Pressure
DASS-21 Depression Anxiety and Stress Scale, 21 item version
EF Executive Functioning
EF CLOX Executive Functioning CLOX task
ISI Insomnia Severity Index
LLT Location Learning Test
NART National Adult Reading Test
OSA Obstructive Sleep Apnoea
PSG Polysomnography
PUCOSA Predicting Usage of Continuous Positive Airway Pressure in
Obstructive Sleep Apnoea
REM Rapid eye movement
TMT Trail Making Test
1
CHAPTER 1: GENERAL INTRODUCTION
Neurocognitive Disturbance in Obstructive Sleep Apnoea: Mechanisms of
Cognitive Harm
Sleep had traditionally been understood as a passive event that occurs in the
absence of alertness (Data & MacLean, 2007). However, as the technology for
measuring changes in physiology has evolved, so too has the understanding of sleep.
Modern science reveals that sleep is an important, dynamic process (Data &
MacLean, 2007). Scalp electrical recordings were first used in the 1930s visually to
identify patterns of brain activity in sleep (AASM, 2007), and in 1967 the first
standardized scoring manual was produced by Rechtschaffen & Kales (AASM,
2007).
The specific physiological and neurological features of each sleep stage can be
measured and recorded using polysomnography (PSG). PSG is the continuous
measurement of the ebb and flow of physiological processes during sleep (Butkov &
Lee-Chiong, 2007). Overnight, laboratory PSG is the gold-standard method for
diagnosing sleep disorders such as obstructive sleep apnoea (Kryger, 2010). During
the overnight sleep study, measurements of oxygen saturation, brain activity, muscle
tone, heart activity and rhythm, breathing rhythm, airflow, eye movements, sound
and gross body movements are obtained. These physiological measurements are used
to stage sleep, and disturbances of sleep.
Human sleep features two broad stages; Rapid Eye Movement sleep (REM), so
named for the characteristic eye movements present during this stage, and Non-
Rapid Eye Movement sleep (NREM) (Kryger, 2010). NREM sleep can be further
broken into 3 sub-stages of sleep; NREM stage 1 (N1), NREM stage 2 (N2) and
NREM stage 3 (N3) (AASM, 2007). Stages N1 and N2 are considered shallow,
Neurocognitive Disturbance in OSA 2
transitory sleep stages, and N3 is considered deep sleep, characterised by slow,
rhythmic brain activity (AASM, 2007; Kryger, 2010). Healthy adults cycle through
each of the four sleep stages in 90-120 minute cycles, totaling approximately 4-6
cycles a night (Carskadon & Dement, 2011). Typically, sleep will start in shallow
stages progressing from N1 to N2 then N3, to REM (Carskadon & Dement, 2011).
However, this pattern is by no means a clean progression, and a person can
experience periods of wake or shallow sleep during or before deeper sleep states or
REM sleep. Sleep stages are classified, or ‘scored’, during PSG by the type of
waveform present in the EEG signal. The waveform present in an epoch of sleep (30
seconds of recorded sleep), and the pattern of waveforms over the course of the night
is known as sleep architecture. Disturbances to sleep are said to disturb sleep
architecture, as they change the waveforms of, transitions between and length of
stages seen in healthy sleep.
Staging adult sleep and scoring events is guided by specific criteria laid out in
The American Academy of Sleep Medicine (AASM) Manual for the Scoring of Sleep
and Associated Events: Rules, Terminology and Technical Specifications (AASM,
2007). This guide provides a comprehensive set of rules for equipment application
and signal evaluation in a PSG. The manual builds on the original scoring manual
written by Rechtschaffen and Kales (Rechtschaffen & Kales, 1968). These
guidelines define not only sleep stages but also sleep disordered events (AASM,
2007).
A sleep-disordered event is the occurrence of an important disturbance to
regular sleep patterns. Such an event may be a pause in breath for greater than 10
seconds (an apnoea), a reduction in breathing (a hypopnoea), regular and periodic
Neurocognitive Disturbance in OSA 3
movement of the legs (restless legs), frequent wakenings (sleep fragmentation), or
other disturbance (e.g. nocturia, nightmares, or movement during REM).
Some of these events can occur several times a night, even in healthy sleepers.
For example, a healthy adult will often wake during the night, many times without
awareness (Sforza et al, 2008). These events become problematic only when they
reach a critical number of times per night, affecting sleep architecture or affecting
daytime functioning (Kryger, 2010). A sleep disorder is a medical disorder
disrupting sleep, and is diagnosed when the number of sleep disrupting events is
above a critical point (Kryger, 2010). Sleep Disordered Breathing (SDB) is a group
of disorders characterised by abnormalities in respiratory patterns during sleep, of
which the most common is Obstructive Sleep Apnoea (OSA) (Al-Lawati, Patel, &
Ayas, 2009).
Adult Obstructive Sleep Apnoea (OSA)
Description, burden of health, risk factors and complications
Adult OSA is a frequent and often under-diagnosed condition characterised by
repeated upper airway (pharyngeal) collapse resulting in sleep fragmentation and
oxygen desaturation. (Butkov & Lee-Chiong, 2007). The condition often manifests
in snoring, nocturnal gasping and choking, excessive daytime sleepiness, un-
refreshed sleep, poor concentration and memory, fatigue, and reduced quality of life
(Al-Lawati et al., 2009; Butkov & Lee-Chiong, 2007; Kryger, 2010). The estimated
prevalence of OSA in the general population is 9% of middle-aged women and 27%
of middle-aged men (Young, Peppard, & Gottlieb, 2002). However, the clinical
prevalence, that is those who experience daytime symptoms and seek assessment
(OSA Syndrome), is between 1-5%, leaving a large proportion of people with OSA
undiagnosed and untreated (Butkov & Lee-Chiong, 2007).
Neurocognitive Disturbance in OSA 4
Untreated OSA is associated with increased healthcare utilization, occupational
injuries, and motor-vehicle accidents (Al-Ghanim, Comondore, Fleetham, Marra, &
Aya, 2008). Hillman, Murphy, Antic, and Pezzulo (2006a) completed a
comprehensive study evaluating the financial cost of sleep disorders in Australia, and
estimated that the total economic burden of sleep disorders in Australia was $US7.5
billion in the year 2004, representing 1.4% of the total burden of disease for
Australia. Sassani et al. (2004) estimate that, for the United States in the year 2000,
there were more than 800,000 motor-vehicle collisions, costing more than $15.9
billion, and claiming more than 14,000 lives, attributable to OSA. These same
authors estimated that these figures could be more than halved if OSA were
appropriately treated (Sassani et al., 2004).
OSA is effectively treated with adherent use of Continuous Positive Airway
Pressure (CPAP) devices. Treatment with CPAP assists, to some extent, with deficits
in cognition, improves quality of life and reduces car accidents (Butkov & Lee-
Chiong, 2007). However, research demonstrates that between 46-83% of people with
OSA are not adherent to the minimum 4 hours a night (Weaver & Grunstein, 2008).
Cognition in OSA
Adequate, undisturbed sleep is theorised to be critical for brain health and
cognitive function. Poe, Walsh, and Bjorness (2010) have reported that increases in
N3 and REM sleep are associated with new learning, and improvements in task
performance. Sleep disturbances such as OSA, disrupt the ability to attain deep sleep
states such as N3 and REM, and are associated with cognitive disturbance (Bucks,
Olaithe, & Eastwood, 2012).
OSA is also associated with disturbances in attention (Findley et al., 1986),
memory and new learning (Bedard, Montplaisir, Richer, Rouleau, & Malo, 1991),
Neurocognitive Disturbance in OSA 5
and executive function (Fulda & Schulz, 2003; Saunamäki, Himanen, Polo, &
Jehkonen, 2010; Saunamäki & Jehkonen, 2007). A recent meta-review (Bucks et al.,
2012) concluded that individuals with OSA show deficits in broadly in attention; in
memory, specifically episodic visual and verbal memory, visuospatial/constructional
abilities; and, in executive function. Cognitive areas that appear to remain unaffected
are language abilities, visual immediate recall, visuospatial learning and
psychomotor functions. This review concluded that there was clear evidence for
dysfunction within the domains of attention and memory, and probable harm in the
domain of executive function, although this has yet to be clarified.
The mechanisms of harm through which OSA causes cognitive dysfunction are
less well understood. The dominant model of neural harm and cognitive dysfunction
in OSA proposes that harm is caused through long-term episodic hypoxia and sleep
fragmentation.
Conceptual models of cognitive harm in OSA
Beebe and Gozal (2002) have proposed a conceptual framework based around
critical roles for sleep fragmentation and nocturnal hypoxia in the development of
cognitive dysfunction in individuals with OSA. In their model, sleep is viewed as a
necessary restorative process for efficient executive functioning and reinforcing
foundations for learning and memory. The nocturnal disturbances of sleep
fragmentation and hypoxia interrupt and corrupt these processes.
Sleep fragmentation. OSA causes sleep fragmentation through the
interruption of sleep by frequent, brief arousals following a respiratory disturbance,
such as an apnoea or hypopnoea (Kryger, 2010). During an apnoea or hypopnea the
individual exhibits partial or full airway obstruction, this restricted flow causes the
person to arouse, which restarts breathing. This can happen many times a night.
Neurocognitive Disturbance in OSA 6
Sleep fragmentation in OSA is proposed to contribute to neurocognitive dysfunction,
specifically decrements in attention (Verstraeten, 2007), and memory (Daurat, Foret,
Bret-Dibat, Fureix, & Tiberge, 2008). Sleep fragmentation contributes to these
dysfunctions by slowing cognitive processing speed (Torun-Yazihan, Aydin, &
Karakas, 2007; Verstraeten, 2007), and disruption of the restorative processes of
sleep (Beebe & Gozal, 2002; Tartar et al., 2010). McKenna et al. (2007) have
demonstrated that sleep fragmentation elevates behavioural, electrographic, and
neuro-chemical measures of sleepiness in rats.
The gold standard treatment in OSA, is with Continuous Positive Airway
Pressure (CPAP) devices. These devices apply gentle air pressure into the pharynx in
order to maintain airway patency (Sullivan, Berthon-Jones, Issa, & Eves, 1981).
Treatment with CPAP improves quality of life and reduces car accidents and, to
some extent, assists with deficits in cognition (Weaver & Grunstein, 2008).
However, between 46-83% of people with OSA use their device for less than 4 of
use hours of use per night (Weaver & Grunstein, 2008). Such low usage may also
result in inadequate treatment of OSA-related cognitive impairment.
In OSA, even with successful amelioration of sleep fragmentation with CPAP
treatment, neurocognitive dysfunction often persists (Beebe, Groesz, Wells, Nichols,
& McGee, 2003). There is evidence of cell death and structural abnormalities in
regions of the brain associated with memory (hippocampus), attention and executive
function (frontal cortices) in animal (Xu et al., 2004) and human models (Zhang,
Lin, Shunwei, Yuping, & Luning, 2011). As such, it seems likely that these deficits
are a primary and direct consequence of OSA (see Beebe and Gozal (2002) for a
review; but see Durmer and Dinges (2005b) for evidence that similar frontal changes
are also seen post sleep loss), rather than as a secondary result of cognitive slowing
Neurocognitive Disturbance in OSA 7
and sleep disturbance. Indeed, one strong proponent of the secondary deficits
argument (Verstraeten, Cluydts, Pevemagie, & Hoffman, 2004), conducted a study
controlling for attention deficits while exploring executive dysfunction differences
between OSA and controls. They concluded that while most of the executive
difficulties were secondary to attention, deficits in the ‘Shifting’ facet of executive
function remained.
Hypoxia. Other authors argue that the deficits remaining after successful
CPAP treatment are caused by the long-term night-time hypoxia experienced by
individuals with OSA (Beebe, 2006). There is a substantial body of literature that
suggests that hypoxia damages the central nervous system through oxidative damage
and contributes to chronic and untreatable cognitive dysfunction (Tsai, 2010).
An oxidant is a toxic substance that can cause oxidative damage to proteins
and lipids within the human body (Butkov & Lee-Chiong, 2007). Peroxidation is the
process by which free radicals or oxidants disrupt the structure of the cellular
membrane of proteins and lipids. Free radicals are produced in normal cellular
respiration, however, in healthy individuals systems have evolved to minimise
oxidative damage and eliminate excess free radicals (Butkov & Lee-Chiong, 2007).
In individuals with OSA these systems either do not function at their fullest capacity
or are burdened by an over-production of free radicals (Butkov & Lee-Chiong,
2007). What results from a long chain of oxidative damage, is damage to cellular
membranes and the production of mutagenic and carcinogenic end products (Xu et
al., 2004). Chronic intermittent hypoxia (CIH) is the long-term cycle of
deoxygenation and reoxygenation of bodily tissues, as is seen in OSA.
CIH is believed to contribute to memory and executive dysfunction seen in
OSA (Beebe & Gozal, 2002). Neuroimaging shows that individuals with OSA
Neurocognitive Disturbance in OSA 8
experience reduced cell neurogenesis and density of the hippocampus (Guzman-
Marin, Bashir, Suntsova, Szymusiak, & McGinty, 2007), the frontal cortex
(Hatipoglu & Rubinstein, 2004), and generalised grey matter reductions (Macey et
al., 2003). Additionally, CIH has been associated with decline in the cognitive
domains associated with these brain regions: memory, attention/speed of processing,
and executive functioning (Beebe, 2006; Beebe et al., 2003).
Summary of proposed mechanisms of harm. Despite strong evidence that
sleep fragmentation and hypoxia have the potential to cause cognitive harm and
dysfunction, and clear evidence that individuals with OSA exhibit cognitive
dysfunction, there remains a dearth of knowledge about the nature of the relationship
between cognitive dysfunction and nocturnal disturbance. Studies have consistently
failed to find a ‘dose-response’ relationship between the severity of OSA (either
hypoxia or fragmentation) and the severity of cognitive deficits found. Indeed, a
search of the literature reveals no empirical test of Beebe and Gozal’s full conceptual
framework, and conflicting results for studies investigating the relationship between
OSA and cognition (Aloia, Arnedt, Davis, Riggs, & Byrd, 2004). This may be for
one of several reasons; 1) use of measures of nocturnal indices of disturbance that do
not separate hypoxia and sleep fragmentation, 2) a lack of theoretically-driven
measurement of cognitive domains, 3) use of problematic statistical methods, or 4) a
failure to account for potentially confounding, inter-individual factors such as
premorbid IQ, sleepiness and age. Each is considered, in turn, below.
Measuring nocturnal disturbance in OSA
People with OSA exhibit both sleep fragmentation and hypoxia, however these
indices are not captured by the primary measure of disease severity in OSA, the
Apnoea Hypopnoea Index (AHI). The AHI is a summation of the total number of
Neurocognitive Disturbance in OSA 9
times an individual experiences an apnoea and/or hypopnoea. However, individuals
can differ on the magnitude and length of desaturation during these events (Kryger,
2010). The AHI makes the assumption that multiple, brief, shallow desaturations are
as problematic for cognition, and other functions affected by OSA (e.g., the
cardiovascular system, and mood), as fewer, deeper, full obstructions leading to
profound oxygen desaturation. People can also exhibit more arousals than marked by
apnoeas and hypopnoeas, since many respiratory disturbances do not meet the
thresholds given by the AASM (AASM, 1999). It is possible that the relative
imprecision of the AHI is the reason that research has consistently failed to find an
association between OSA severity (indexed by the AHI) and cognitive dysfunction
(Aloia et al., 2004).
The studies that have considered sleep fragmentation or hypoxia in isolation
suggest differential harm. This idea is captured in both conceptual models (Beebe,
2006; Beebe et al., 2003), and research. There is evidence to suggest that attention
may be more impaired by sleep fragmentation (Verstraeten, 2007), whilst hypoxia
greatly affects memory and executive functioning (Beebe & Gozal, 2002). This
proposal is tentative at best, and needs further examination. A review by (Aloia et
al., 2004) examined the relationship between hypoxia and sleep fragmentation and
domains of cognitive dysfunction, across 37 peer-reviewed papers, of which 11 had
examined the relationship between some measure of nocturnal disturbance and a
cognitive domain. The review revealed that approximately half the papers reported a
relationship between attention and hypoxia (5 of 10 papers) and, attention and sleep
fragmentation (6 of 9 papers), and less than half reported a relationship between
executive function or memory and hypoxia (3 of 10 and 2 of 8 papers, respectively)
and/or sleep fragmentation (2 of 9 and 2 of 7 papers, respectively). The authors
Neurocognitive Disturbance in OSA 10
concluded that no relationship could be established between the degree of cognitive
dysfunction and the extent of hypoxia and sleep fragmentation. However, they
highlighted that there were many issues with this conclusion. For instance, the many
different papers utilized a variety of nocturnal disturbance measures which may not
measure the same aspect of disturbance. For example in calculating sleep
fragmentation Aloia et. al (2004) necessarily combined measures of total arousal
(ArI) with global measures of disturbance (AHI). The authors also aggregated many
different cognitive assessments across studies (more on this point below). The
authors concluded by saying their results were equivocal, and more studies needed to
be conducted to elucidate the link between cognition and mechanisms of harm in
OSA.
Despite no clear link between cognition and mechanisms of harm from meta-
analyses and laboratory based studies neuroimaging studies do show that sleep
disturbance, hypoxia in particular, visibly affects the brain (Canessa et al., 2011).
This harm occurs through similar mechanisms to those that cause stress and damage
in the cardiovascular system (Hamilton & Naughton, 2013), without the usual
protective mechanisms such as vasodilation, probably due to the cyclic nature of
hypoxia in OSA (Kato et al., 2000).
Despite neuroimaging and experimental evidence that sleep fragmentation and
hypoxia do cause harm to brain structures and cognitive function, a lack of clarity
remains in the cognitive profile of individuals with OSA, and the relationship
between cognitive dysfunction and mechanisms of harm. This may be due to several
reasons, of which 3 are commonplace in the literature; 1) a lack of theoretically
sound division and measurement of cognition domains, 2) use of problematic
Neurocognitive Disturbance in OSA 11
statistical methods, and 3) no measurement of inter-individual differences such as
age or Premorbid IQ that can moderate the impact of OSA on cognition.
Measurement of cognition in OSA
Further complicating an understanding of the dose-response relationship
between nocturnal harm and cognition is the wide range of tests used in different
research papers. Researchers have utilised an array of tasks to measure whole
domains of cognition such as memory or executive function. However, these
different tasks likely capture different facets of the domain. For example, executive
function in OSA has been assessed by tasks as diverse as the Trail Making Test (a
test of the ability to join numbers and letters in alternating sequence, as quickly as
possible, 1 – A – 2- B – 3 –C and so on), and verbal fluency (a test of the ability to
generate words beginning with a letter, e.g. F, A. and S) (Saunamäki & Jehkonen,
2007). Each of these tasks captures a different facet of executive function: set-
shifting or cognitive flexibility and generativity, respectively. A review of the impact
of OSA on episodic memory by Wallace and Bucks (2012) highlighted the need to
separate cognitive domains into sub-domains. Wallace and Bucks (2012) showed
that the discrepant findings in episodic memory testing may have been a function of
deficits in some domains of memory (verbal episodic memory (immediate recall,
delayed recall, learning and recognition) and visuo-spatial episodic memory
(immediate and delayed recall)), but not visual immediate recall or visuo-spatial
learning.
Likewise, an examination of the many tasks used to examine executive
function, theoretically divided, may explain some of the discrepant findings within
the literature. Chapter 2 reports the first meta-analysis of executive deficit in OSA
that considers this issue in detail.
Neurocognitive Disturbance in OSA 12
Use of problematic statistical methods
Many examinations of the relationship between cognition and nocturnal
disturbance in OSA have used small sample sizes (Beebe et al., 2003) and traditional
regression techniques. These regression techniques are unable to capture individual
variation and measurement error (Byrne, 2010). This lack of precision may
contribute to the differential findings in the literature.
In order to account for measurement error and individual variation the present
thesis used Structural Equation Modelling (SEM). SEM is an extension of multiple
regression designed to test a set of hypothesised relationships between variables,
estimated simultaneously (Ullman & Bentler, 2012). It provides a mechanism
through which to examine relationships between hypothesized constructs, whilst
controlling for individual differences, such as premorbid intelligence and sleepiness,
and accounting for measurement error.
SEM is particularly useful when examining hypothesized constructs such as
memory, attention, and executive function, as it can account for measurement error
that naturally exists between the ‘pure’ construct and its measurement, providing a
stringent test of the latent structure (Byrne, 2010; Iaccobucci, 2010; MacCallum &
Austin, 2010). In addition, it allow the simultaneous exploration of the impact of
OSA (through hypoxia and sleep fragmentation) on multiple aspects of cognition
(i.e. memory, attention and executive function), thus reducing the risk of a Type 1
error which would otherwise arise from testing the impact on each cognitive
outcome variable individually (Byrne, 2010).
Neurocognitive Disturbance in OSA 13
Inter-individual differences moderating the impact of OSA on cognition
Premorbid cognitive functioning. Pre-morbid cognitive functioning alters
the neurocognitive expression of OSA (Tsai, 2010). Cognitive reserve is the concept
that a high level of pre-morbid cognitive ability acts to 'buffer' the effect of
neurocognitive trauma (LaRue, 2010). Alchanatis et al. (2005) reported that high-
intelligence (an index of cognitive reserve) participants with OSA had the same
attention and alertness patterns as high-intelligence participants without OSA.
However, normal-intelligence participants with OSA experienced decline in
attention and alertness compared to normal-intelligence controls. These authors
theorised that high-intelligence protected participants from the negative impact of
OSA on cognition.
Subjective sleepiness. Individual differences in subjective sleepiness in OSA
may modify the way an individual performs on neurocognitive tests (Alchanatis,
Zias, Deligiorgis, Liappas, et al., 2008). This may be because high sleepiness lessens
the ability to direct cognitive resources, and attend to the task at hand. Consistent
with this view, sleepy individuals perform less well on neurocognitive tasks, than
non-sleepy individuals (Durmer & Dinges, 2005b; Naismith, Winter, Gotsopoulos,
Hickie, & Cistulli, 2004).
Age. Participant age is an important variable in research exploring OSA and
cognition. OSA is related to age in two ways. First, the risk of having OSA increases
with increasing age (Alchanatis, Zias, Deligiorgis, Chroneou, et al., 2008). Indicating
that the older the individual the more likely they are to have OSA. Secondly, OSA is
commonly undiagnosed (Young, Evans, Finn, & Palta, 1997) as it can only be
Neurocognitive Disturbance in OSA 14
diagnosed with an overnight sleep study. This means that older people may have had
undiagnosed OSA for a longer period of time, leading to larger cumulative deficits.
Our best estimate of disease duration comes from subjective reports of snoring or
demographic risk factors, such as obesity (Marcus, Pothineni, Marcus, & Bisognano,
2014) of which the individual (or partner) may or may not be aware, and which may
or may not indicate OSA (Cirignotta et al., 2009).
Summary and contents of chapters
OSA causes cognitive dysfunction. Reviews and meta-analyses have clarified
the pattern of deficits in attention, and memory, however the pattern of executive
dysfunction remains unclear. Hypoxia and sleep fragmentation are the hypothesised
mechanisms of harm in OSA, as posited by published conceptual models, however
these models and the relationship between nocturnal disturbance (hypoxia and sleep
fragmentation) and attention, memory and executive dysfunction have yet to tested.
This thesis aimed to examine and clarify the relationship between cognition,
and sleep disturbance in OSA. In particular it examined the relationship between the
broad domains of cognition shown to be disordered in OSA: attention, memory, and
executive function, and nocturnal disturbance: sleep fragmentation and hypoxia.
In order to do this, first it was necessary to clarify the pattern of executive
function deficits. The results of a meta-analysis of executive dysfunction in OSA
before and after treatment are presented in Chapter 2. This study was published in
the journal Sleep (2013); 36 (9); 1297-305.
Detailed information about recruitment, participants, materials and
methodology for Chapters 4 and 5 are presented in Chapter 3. As Chapters 4 and 5
Neurocognitive Disturbance in OSA 15
were submitted for publication, and due to journal word restrictions, these chapters
have little detail on recruitment, participants, materials and methodology.
Premorbid IQ, age and daytime sleepiness, independently affect cognitive
performance, and there is a need to control for these inter-individual factors when
examining the relationships between cognition and OSA (Alchanatis et al., 2005;
Alchanatis, Zias, Deligiorgis, Chroneou, et al., 2008). Age can be quantified, and
premorbid IQ can be estimated using psychometrically validated tools (Nelson &
Willison, 1991), however, the factor structure of the most widely-used measure of
daytime sleepiness, the Epworth Sleepiness Scale (ESS: used in this thesis) has been
questioned (Smith et al., 2008). Factor analyses of the ESS have revealed 1, 2, or 3
possible underlying constructs. In order to understand how best to utilise the ESS,
Chapter 4 examines the factor structure of the ESS in community and clinical
samples using confirmatory factor analysis. This chapter was published in Sleep and
Breathing (2012); 17(2); 763-9.
The final, investigative chapter used the findings from the meta-analysis and
investigation of the ESS, in structural equation models examining the relationships
between attention, memory and executive function, and hypoxia and sleep
fragmentation, accounting for intelligence, age and sleepiness. This chapter has been
submitted to Sleep and Breathing for review as of the 17.12.2013.
The final chapter of the thesis, Chapter 6, presents a general discussion of the
findings, strengths, original contributions, and future directions.
Neurocognitive Disturbance in OSA 16
CHAPTER 2: EXECUTIVE FUNCTION IN OSA
Preface
Cognition is affected by OSA, however, until recently the literature has been
divided as to the profile of cognitive dysfunction in OSA. Recent reviews have
summarised which aspects of attention and memory are impacted by OSA. However,
it remains unclear what specific aspects of executive function are impaired.
Chapter 2 presents a meta-analysis reviewing the literature regarding which
specific aspects of executive function are impaired in OSA and improved by CPAP.
This chapter was published in the journal Sleep: Olaithe, M. and Bucks, R.S.
(2013). Executive Dysfunction in OSA Before and After Treatment: A Meta-
Analysis. SLEEP, 36(9); 1297-305. It is presented below, as published, but
formatted for consistency with the rest of the thesis.
Neurocognitive Disturbance in OSA 17
Abstract
Study Objectives: Obstructive Sleep Apnoea (OSA) is a frequent and often under-
diagnosed condition that is associated with upper airway collapse, oxygen
desaturation, and sleep fragmentation leading to cognitive dysfunction. There is good
meta-analytic evidence that sub-domains of attention and memory are affected by
OSA. However, a thorough investigation of the impact of OSA on different sub-
domains of executive function has yet to be conducted. This report investigates the
impact of OSA and its treatment, in adult patients on five, theorised, sub-domains of
executive function.
Design: An extensive literature search was conducted of published and unpublished
materials, returning 35 studies that matched selection criteria. Meta-analysis was used
to synthesise the results from studies examining the impact of OSA on executive
functioning compared to controls (21 studies) and before and after treatment (19
studies); 5 studies met inclusion in both categories.
Measurements: Research papers were selected which assessed five sub-domains of
executive function: Shifting, Updating, Inhibition, Generativity and Fluid Reasoning.
Results: All 5 domains of executive function demonstrated medium to very large
impairments in OSA independent of age, and disease severity. Furthermore, all sub-
domains of executive function demonstrated small to medium improvements with
CPAP treatment.
Discussion: Executive function is impaired across all five domains in OSA, these
difficulties improve with CPAP treatment. Age and disease severity did not moderate
Neurocognitive Disturbance in OSA 18
the effects found, however, further studies are needed exploring the extent of primary
and secondary effects, and the impact of age and pre-morbid ability (cognitive
reserve).
Neurocognitive Disturbance in OSA 19
Executive Dysfunction in OSA Before and After Treatment: A Meta-Analysis
Obstructive sleep apnoea (OSA) is a frequent and often under-diagnosed
condition that is associated with upper airway collapse, oxygen desaturation, and
sleep fragmentation leading to sleepiness, hypertension, increased risk of cardiac
disease, and neurocognitive disturbance. (Al-Lawati et al., 2009; Butkov & Lee-
Chiong, 2007; Young, Palta, & Dempsey, 1993) Untreated OSA is associated with
increased healthcare utilization, occupational injuries, motor-vehicle accidents (Al-
Ghanim et al., 2008; Hillman, Murphy, Antic, & Pezzulo, 2006b; Sassani et al.,
2004) and neurocognitive sequelae in memory, attention and executive function.
(Butkov & Lee-Chiong, 2007; Tsai, 2010) The gold standard treatment for OSA is
Continuous Positive Airway Pressure (CPAP). (Kryger, 2010; Weaver & Grunstein,
2008)
To date, most reviews of cognitive functioning in OSA have inspected
cognition as a whole, collapsing research findings into ‘memory’, ‘executive
function’ or ‘attention’ domains (for example see, (Aloia et al., 2004; Beebe et al.,
2003)). As the evidence base grows, however, it is both possible and desirable to
explore the cognitive burden of OSA within subcomponents. Based on current
neurocognitive theory, there are functional and biological grounds for segregating
cognitive domains or functional systems into such subcomponents. (Adrover-Roig,
Sesé, Barceló, & Palmer, 2012; Lezak, Howieson, & Loring, 2004) These
subcomponents work in concert to produce what we colloquially know as memory,
attention and executive function. (Elliot, 2003; Larner, 2008)
Recently, Wallace and Bucks (Wallace & Bucks, 2012) divided episodic
memory into theoretically-driven subcomponents, revealing deficits in individuals
with OSA in verbal episodic memory (immediate recall, delayed recall, learning and
Neurocognitive Disturbance in OSA 20
recognition) and visuo-spatial episodic memory (immediate and delayed recall), but
not visual immediate recall or visuo-spatial learning. This theoretically-driven
division of memory reveals that not all components of memory are dysfunctional in
OSA and provides an explanation for the mixed findings in this field. A similar
approach might prove fruitful when exploring executive dysfunction in OSA. In a
review, Saunamaki et al. (Saunamäki & Jehkonen, 2007) demonstrated that aspects
of executive function may also be impaired or preserved in OSA. They divided
executive functioning by test, demonstrating deficits in Digit Span Forwards, Corsi
Block Tapping task, Double encoding task, Wisconsin Card Sorting Test, Phonemic
fluency, Rey-Osterreith Complex Figure test, and Maze tasks. However, by meta-
analysing the data by test, this review did not aggregate executive functions using a
theoretical framework. In addition, Saunamaki et al. included some tests that do not
primarily measure executive function (i.e. Digit Span Forwards, Rey-Osterrieth
Complex Figure test, the Trail Making Test Part A and the Corsi Block Tapping
task), making it difficult to determine which subcomponents of EF, mapped by
which tests, are impaired in OSA.
Executive function is an individually controlled and conscious effort to guide
the operation of various cognitive processes and thereby regulate cognition. (Banich,
2009; Elliot, 2003; Funahashi, 2001; Lezak et al., 2004; Miyake, Friedman,
Emerson, Witzki, & Howerter, 2000; Suchy, 2009) Like other cognitive domains,
executive function is multidimensional. (Chan, Chen, Cheung, Chen, & Cheung,
2006; Lezak et al., 2004; Miyake et al., 2000; Suchy, 2009) Miyake et al., (Miyake et
al., 2000) Fisk & Sharp (Fisk & Sharp, 2004) and Adrover-Roig et al. (Adrover-Roig
et al., 2012) present an empirical basis for specifying how executive functions are
organised, and what roles different subcomponents play. Miyake et al. (Miyake et
Neurocognitive Disturbance in OSA 21
al., 2000) divide executive functioning into (a) Shifting between tasks or mental sets,
(b) Updating and monitoring of working memory representations and (c) Inhibition
of dominant or pre-potent responses. (Lehto, Juuja¨rvi, Kooistra, & Pulkkinen, 2003)
Fisk and Sharp (Fisk & Sharp, 2004) and Adrover-Roig et al (Adrover-Roig et al.,
2012) utilized this same 3-factor structure, but proposed a fourth component;
efficiency of access to long term memory (called Generativity in the present report,
for brevity). This four-component structure has been confirmed in multiple
populations with factor analysis (exploratory and confirmatory) (Adrover-Roig et al.,
2012; Fisk & Sharp, 2004; Lin, Chan, Zheng, Yang, & Wang, 2007; Montgomery,
Fisk, Newcombe, & Murphy, 2005).
Lezak, Howieson and Loring (Lezak et al., 2004) and Strauss, Sherman and
Spreen (Strauss, Sherman, & Spreen, 2006) define a set of tasks that do not tap
executive function per say but rather an overarching system of reasoning and
problem solving. These tasks involve complex, higher order abstraction, problem
solving and concept formation and include tasks such as Porteus Mazes or Clock
drawing tasks. (Lezak et al., 2004; Strauss et al., 2006) The four-factor model
defined above does not account for such tasks; however they abound in OSA
literature on executive function and are considered a part of executive functioning in
neuropsychological theory (Lezak et al., 2004; Strauss et al., 2006). Hence, in the
present paper a class of executive function tasks, called Fluid Reasoning, was created
to capture this concept.
The present paper builds on past reviews and meta-analyses examining
executive functioning in adults with OSA within current neuropsychological theory
of executive function. No previous meta-analysis in OSA has assessed EF
dysfunction, or the effect of treatment, within these 5, theoretically-motivated
Neurocognitive Disturbance in OSA 22
domains: Shifting, Updating, Inhibition, Generativity, and Fluid Reasoning. We
addressed three questions: 1) which specific executive functions are affected by the
presence of untreated OSA?; 2) if executive functions are impaired, does treatment
help to remediate these deficits?; and 3) are any of these effects moderated by
publication status, sample source, study design, age, disease severity, or control
screening?
Method
Search Strategy
Data for this meta-analysis consisted of empirical articles published in peer-
reviewed journals over the past 24 years (Jan 1987 – Nov 2011). Details of the
search methodology employed are outlined in Figure 1.
Neurocognitive Disturbance in OSA 23
Search terms:(Apnoea OR sleep-disordered breathing) AND (Cognition OR Cognitive ability OR Mental Status OR
Neuropsychology OR Memory OR Attention OR Vigilance OR Executive OR Psychomotor)
Data Analysis:
Calculated effect sizes
Calculated statistical heterogeneity
Publication bias
Full text copies retrieved for evaluation (n =
269) using quality assessment criteria
Studies excluded (n = 234) Reasons:
1. Paper was not in an appropriate format (e.g. review article)
2. The article was not in English
3. Participants were <18 years
4. Participants did not have primarily obstructive sleep apnoea
5. Did not assess EF
6. Test used was inadequately described
7. Test did not have acceptable validity and/or reliability
8. Data were presented in such a way that effect sizes could not
be calculated
9. PSG not done on participants
10. PSG conducted >12months before/after cognitive assessment
11. Special population was used (e.g. OSA in TBI or insomnia)
12. Data were published elsewhere
13. Group matching was inappropriate (e.g. IQ higher in one
group than the other)
Abstracts excluded (n = 177) Reasons:
1. Participants did not have primarily obstructive sleep apnoea
2. Paper did not examine executive function
3. Participants were <18 years
4. The article was not in English
5. Paper was not in an appropriate format (e.g. review article)
Extracted descriptive data (n = 35 (34 + 1
personal communication)): author/s, publication
status, year of publication, study design, sample
size, participant details, co-morbidities screened
for, source of OSA sample and executive
function assessments employed.
Electronic Databases searched (Keyword and MeSH explode): Medline R (n = 463), Psych Info (n = 127), PubMed (n =
1757), EMBASE (n = 771), CINAHL (n = 118), CCTR (n = 31), NHS EED (n = 47)
Grey Literature: SIGLE (n = 15), NTIS (n = 1)
Conference proceedings: Conference proceedings citation index science (n = 212)
Dissertations and Theses: Proquest dissertations and theses (n = 71)
Internet: Google scholar (n = 8,070)
Handsearching: Index Medicus, Exerpta Medica, References of included articles (n = 60), Contact with authors of
unpublished or studies with incomplete data (n = 33)
11,302 duplicates and articles not relevant to topic removed
Titles and abstracts screened (n = 446)
Figure 1. Flow chart of study inclusion and exclusion
Neurocognitive Disturbance in OSA 24
An extensive computer assisted literature search was conducted using
electronic databases (Keyword and MeSH explode) for published articles (Medline
R, PsychInfo, PubMed, EMBASE, CINAHL, CCTR, NHS EED), grey literature
(SIGLE, NTIS), conference proceedings (Conference Proceedings Citation Index:
Science), dissertations and theses (Proquest Dissertations and Theses), the Internet
(Dogpile, Omni Medical search engine, Mednet) and via handsearching (Index
Medicus, Exerpta Medica, references of included articles, contact with authors of
unpublished studies). Unpublished studies were included in the search, to avoid
publication bias.
The terms ‘apnoea OR sleep-disordered breathing’ were combined with
‘Cognition OR Cognitive ability OR Mental Status OR Neuropsychology OR
Memory OR Attention OR Vigilance OR Executive OR Psychomotor’. The terms
chosen covered a wide range of cognitive functions to capture tests that had been
mislabelled or utilised to measure other cognitive domains.
Additional relevant articles were retrieved from the reference lists of studies
included in the original search, conference proceedings and dissertations.
Furthermore, key authors who have published articles on the relationship between
OSA, cognition and CPAP treatment were contacted asking if they were aware of
any other relevant published or unpublished studies.
Study selection criteria
This review included studies that assessed executive function in adults with
OSA as defined by an Apnoea Hypopnoea Index (AHI) of > 5 per hour of sleep.
(AASM, 1999) In all instances, except one, studies were excluded if OSA
participants were not diagnosed using overnight polysomnography and/or if they did
not include a control sample, if group matching was inappropriate (e.g. IQ
Figure 1 Flow chart of search, retrieval and inclusion process.
Neurocognitive Disturbance in OSA 25
statistically different between control and OSA groups) or there were no baseline
data (participants were assessed after treatment only). The exception to this rule was
Antic et al., (N. Antic, September, 2012; N. A. Antic et al., 2011) where participants
were administered overnight oximetry instead of PSG. This paper was included as
the oximetry was validated against in-laboratory PSG in a random selection of 50%
of the participants.
Additionally, papers were excluded if the PSG was conducted more than 12
months before/after neuropsychological profiling was completed. These studies were
excluded as individuals may lose or gain weight, or change their lifestyle habits (e.g.
drink, smoke or exercise more or less) which may alter the severity of their OSA.
(Cowan & Livingston, 2012; Ong, O’Driscoll, Truby, Naughton, & Hamilton, 2012)
This review considered only studies with adult participants (≥18 years), not
from special populations (e.g. people with Down’s syndrome, insomnia, or traumatic
brain injury). Research demonstrates that there are etiological differences between
adult and childhood OSA (Cheng, Dai, Wu, & Chen, 2012) thus the latter was not
addressed here.
The present review aimed to delineate the pattern of executive deficits in OSA,
hence studies that included a majority of central or mixed sleep apnoea patients were
excluded. Research demonstrates that the pathophysiology, epidemiology, and
clinical characteristics of central sleep apnoea and OSA are distinct. (T Young, P.E
Peppard, & D. J. Gottlieb, 2002)
Papers were excluded if the tests used were inadequately described such that
acceptable validity and/or reliability could not be confirmed, the paper was a review
paper not a study, if it reported data already included in the present review (in this
instance the most complete data set was selected) or was a cross-over trial. Cross-
Neurocognitive Disturbance in OSA 26
over trials were excluded as research does not provide any definitive information
regarding length of washout period required. (Phillips et al., 1990; Sullivan et al.,
1981; Sullivan & Issa, 1985)
The present study was only able to examine CPAP treatment, as after
evaluating studies with the exclusion criteria, there remained an insufficient number
of other treatment studies (No oxygen therapy, positional therapy, drug trial or
weight loss studies, 1 surgical study, 3 Mandibular Advancement Splint (MAS)
studies, 3 studies with mixed treatments). Nor did the present review examine
medication studies as these (e.g. modafinil and ardmodafinil) may alter alertness,
cognitive function and judgement without treating underlying nocturnal symptoms.
(Estrada, Kelley, Webb, Athy, & Crowley, 2012; Ray et al., 2012) Although research
demonstrates that these medications can be helpful in conjunction with CPAP where
there is residual sleepiness, the present study aimed to look at the effect of OSA on
cognitive function, and such medications may have confounded these results.
Furthermore, studies were not considered if the data were presented in such a
way that effect sizes could not be calculated even after contact with the author. We
contacted 32 authors for further details on 33 research papers. Five authors or their
representatives replied. Of these, 2 authors were deceased, 1 had no more detail to
provide, and 2 emailed further data. Data received from N. Antic (N. Antic,
September, 2012) was utilised in the present meta-review. We also received further
data from M. Barnes, (Barnes et al., 2002) however this was later excluded as it was
from a cross-over trial.
Given that it is difficult to keep participants and experimenters blinded to
group (OSA or Control, Treatment or No treatment) in OSA studies when assessing
Neurocognitive Disturbance in OSA 27
neuropsychological function, (Redline et al., 1997) this review did not exclude
unblinded studies.
Finally the present paper included studies in which controls were screened
using PSG or with questionnaires. Despite the risk of undetected OSA in the control
sample, (Sharwood et al., 2012) evidence from Wallace and Bucks (Wallace &
Bucks, 2012) suggested that comparing OSA participants with controls within
memory domains, with and without PSG screening of controls, did not dramatically
reduce the significance of the effects found, and would have reduced the number of
studies available per subcomponent. Rather, the present meta-analysis considered
control screening method as a moderator instead.
Quality Assessment
The authors (MO, RSB) independently reviewed articles according to the
selection criteria. Where there were disagreements about whether or not to include
an article, the authors discussed and came to an agreement.
The data for all included studies were extracted and coded by the first
author. The second author extracted and coded the data for 10 randomly selected
studies. The intra-class correlation coefficient between the data extracted by the first
and second author was r = 0.99 (CI: 0.99-1.00).
Study Categorization
Included studies (N = 35; 34 studies + 1 personal communication) were
divided into two non-exclusive categories, (1) comparisons of pre-treatment OSA
groups to controls were used to identify the specific pattern of executive dysfunction
present in untreated OSA (n = 21), and (2) CPAP treatment efficacy studies were
used to establish whether executive impairments were permanent in OSA (n = 19): 5
studies met criteria for inclusion in both groups.
Neurocognitive Disturbance in OSA 28
Categorisation of Executive Function
Table 1 presents the sub-domains of executive function and the tests ascribed
to these sub-domains.
Table 1.
Sub domains of executive function and the tests accessing each sub-domain.
Category Description Tests that tap this cognitive skill
Shifting Shifting back and forth between multiple tasks, operations
or mental sets. Requires the disengagement of an irrelevant
task set and subsequent engagement of a relevant task set
when a new operation must be performed on a set of stimuli,
necessary to overcome proactive interference or negative
priming due to having recently performed a different
operation.
- Wisconsin Card Sorting Test1 (18)
- Trails B (3)
- Switching task (26)
Updating Updating and monitoring of working memory
representations. Requires the monitoring and coding of
incoming information for relevance to the task, and then
appropriately revising items held in working memory by
replacing old, no longer relevant information with new more
relevant information. Dynamically manipulate the contents
of working memory.
- N-back tasks (24)
- Digit span backwards1 (9)
- WAIS-R Arithmetic (13)
- ANAM Mathematical processing
(14)
- ANAM running memory (17)
Inhibition Inhibition of prepotent, dominant or automatic responses
when necessary. An internally generated act of control.
- Towers1 (19)
- Stroop task (6)
- Go No-go task (4)
Generativity Speed and efficiency of access to long-term memory. An
independent ability to create, generate or produce content
without any input from what or whom?
- Verbal fluency tasks1 (2)
Fluid reasoning Concept formation/abstraction & problem solving tasks. An
intentional cognitive process that does not occur
automatically, but rather involves the use of deliberate and
controlled mental actions to solve novel problems.
- Mazes2 (12)
- Ravens progressive matrices2 (1)
- Picture completion2 (22)
- WAIS-R Picture arrangement2 (11)
- WAIS-R Block design (8)
- Stockings of Cambridge (23)
Neurocognitive Disturbance in OSA 29
Note. 1 = mapping of tests to category as recommended by Miyake (2000), Fiske & Sharp (2004); Adrover-Roig et al
(2012) 2 = mapping of tests to category as recommended by Lezak, Howieson and Loring (2004); Strauss, Sherman and Spreen
(2006): Numbers in parentheses relate to which tests were used in individual articles, for more details see Appendix 1.
Data Extraction and Analysis
Data extracted and coded from the final articles included author/s, whether
published or not, journal and year of publication (if applicable), study design, sample
size and participant details when available (gender, years of formal education, body
mass index (BMI), age, diagnostic criteria, AHI or RDI, oxygen desaturation indices
[time spent below 90%: CT90] and sleep fragmentation indices [Arousal Index:
ArI]), source of OSA sample (clinical vs. community) and neuropsychological
assessments employed (See Appendix 1 for further details). Means, standard
deviations and sample size were extracted to examine the relationships between the
variables of interest.
In the instance that participants had been assessed at multiple time points after
CPAP treatment we chose the most distant time point, as we wanted to examine the
effect of continuous treatment on executive dysfunction. Furthermore, if participants
were divided into compliant (> 4hrs for 80% of nights) and non-compliant users,
only the compliant user information was included. These two decisions were made
so as best to evaluate the benefit to individuals who utilise their CPAP devices as
- WAIS-R Similarities2 (7)
- Object assembly (10)
- Clocks (20)
- Twenty questions2 (21)
- Category tests2 (5)
- ANAM Matching to sample (15)
- ANAM logical relations (16)
- Five point design task (25)
Neurocognitive Disturbance in OSA 30
recommended over the long term. These choices led us to lose only 1 subsample
from a paper, but no whole papers.
Data Processing
The program Comprehensive Meta-Analysis version 2.2.064 (Borenstein,
Hedges, Higgins, & Rothstein, 2005) was used to synthesize data, calculate effect
sizes, and create Forest plots.
Results
Description of Studies
From the articles identified (N = 35), 21 studies compared people with
untreated OSA to a control sample, 19 compared people with OSA before and after
treatment; 5 studies had both a comparison to controls participants and
neurocognitive testing performance before and after treatment. These studies
represent 40 samples. Only 1 study was recruited from a community setting, (Quan
et al., 2006) thereby making 98% of studies from clinical settings.
In total, there were 551 healthy controls (74% male; Mean age 49.46±8.96
years; Mean ESS 5.52±2.41; Mean BMI 25.42±2.50) and 1010 participants with
OSA (81% male; Mean age 50.40±7.43 years; Mean AHI/RDI 47.58±15.98; Mean
Arousal Index (ArI) 36.15±17.66; Mean Cumulative Time below 90% oxygen
saturation (CT90) 40.07±28.55; Mean education 13.73±1.48years; Mean months of
CPAP treatment 2.89±2.22 months; Mean Hours CPAP use per night 5.34±1.01;
Mean Epworth Sleepiness Score (ESS) 12.02±2.38; Mean Body Mass Index (BMI)
33.12±2.76). Individual study details for sample size, publication source, age, indices
of disease severity (AHI or RDI), oxygen desaturation (CT90) and sleep
fragmentation (ArI), years of education, and length of CPAP treatment are given in
Appendix 1.
Neurocognitive Disturbance in OSA 31
In the present meta-analysis only studies with matched control and OSA
participant variables were chosen, except that as expected, the control and OSA
groups differed significantly in BMI, t(1228) = -56.03, p < 0.001, and ESS, t(1118) =
-51.15, p < 0.001 scores.
Calculation of Effect Sizes
Random effect sizes were calculated. The random effects model assumes that
each study has a different underlying ‘true’ effect size due to differing sample
demographic variables. (Rosenthal, 1995) In the present meta-analysis, samples
differed on such variables as disease severity, age, gender, screening measures,
oxygen saturation and sleep fragmentation (See Appendix 1). A random effects
model accounts for these between-studies differences, as well as within-study
participant differences.
An effect was calculated for each sample across each of the five domains. For
comparisons between the OSA group and healthy controls, Cohen’s d was calculated
according to the formula 𝑑 =𝑚𝑒𝑎𝑛 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠 −𝑚𝑒𝑎𝑛 𝑂𝑆𝐴
𝑝𝑜𝑜𝑙𝑒𝑑 𝑆𝐷, where effect sizes of d ≤ 0.20
are considered small, d = 0.50 medium, d ≥ 0.80 large and d ≥ 1.00 very large.
(Cohen, 1977) Larger, positive effects indicate poorer performance for the OSA
group.
In OSA group pre-treatment compared to post-treatment, Cohen’s d effect
sizes were calculated with the formula = 𝑚𝑒𝑎𝑛 𝑝𝑜𝑠𝑡−𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡−𝑚𝑒𝑎𝑛 𝑝𝑟𝑒−𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡
𝑝𝑜𝑜𝑙𝑒𝑑 𝑆𝐷.
Higher scores indicate greater improvement post CPAP treatment.
The random effect size estimates between OSA and control groups for each
domain are displayed in Table 2. All five sub-domains of executive function were
impaired, compared to controls. A very large effect was noted for Inhibition, a large
Neurocognitive Disturbance in OSA 32
effect was present for Updating and Fluid Reasoning, and medium effect sizes were
present for Shifting and Generativity. These results indicate medium to very large
deficits in executive function performance across all five domains in individuals with
OSA, when compared to control.
The random effect size estimates between OSA pre-treatment and post-
treatment are displayed in Table 2. All five sub-domains of executive function
demonstrated improvement after CPAP treatment. Medium effect sizes were found
in Shifting and Inhibition, and small effect sizes were noted in Updating, Fluid
Reasoning and Generativity. These results indicate small to medium size
improvements across all five domains of executive function with CPAP treatment.
Neurocognitive Disturbance in OSA 33
Table 2.
Mean effect sizes for the differences between OSA to control groups, and pre-treatment to post-treatment groups
Domain Effect Size Statistics Homogeneity Statistics N d 95% CI Z p Q (df) p Tau I2
Lower Upper OSA to controls Shifting 15 0.53 0.38 0.92 4.78 < 0.001 41.68(14) < 0.001 0.42 66.41 Updating 14 0.91 0.49 1.32 4.30 < 0.001 71.12(13) < 0.001 0.71 81.72 Inhibition 9 1.12 0.55 1.69 3.83 < 0.001 57.17(8) < 0.001 0.79 86.01 Generativity 8 0.59 0.34 0.85 4.58 < 0.001 8.25(13) 0.311 0.14 15.18 Fluid Reasoning 11 0.80 0.42 1.19 4.11 < 0.001 44.12(10) < 0.001 0.55 77.33 Pre to post CPAP treatment Shifting 14 0.66 0.31 1.00 3.75 < 0.001 47.12(13) < 0.001 0.58 72.41 Updating 10 0.46 0.15 0.77 2.90 0.021 15.94(9) 0.068 0.33 43.55 Inhibition 9 0.57 0.31 0.86 4.17 < 0.001 10.75(8) 0.216 0.21 25.57 Generativity 8 0.33 0.13 0.53 3.27 0.001 5.21(7) 0.635 0.00 0.00 Fluid Reasoning 10 0.37 0.18 0.56 3.85 < 0.001 10.02(9) 0.349 0.10 10.27
Forest plots for individual studies are available in Appendix 1.
Neurocognitive Disturbance in OSA 34
Heterogeneity
Heterogeneity of effect sizes is a measure of difference between a study’s true
effect size and the observed effect size. The true effect size is the actual effect size in
the underlying population, whilst the observed effect size is the effect measured in
the sample. (Borenstein, Hedges, Higgins, & Rothstein, 2009) Heterogeneity was
investigated visually with Forest plots, and statistically using Cochrane’s Q statistic,
the T2 and I2 statistics (See Tables 2 and 3, and Appendix 1, for individual Forest
plots). When the Q statistic is significant, this suggests there is a significant
difference between the observed and true effect. However the Q statistic is
vulnerable to small sample size, hence T2 and I2 can provide an estimate of the
proportion of real variance caused by extraneous study variables such as age or test
used. (Borenstein et al., 2009) In any instance that Q was significant, I2 was
examined to quantify the degree of heterogeneity.
For the comparisons between controls and individuals with OSA, significant
heterogeneity was present in all domains. However, the I2 ranged between 77.33 and
86.01 suggesting at least 77% of the variance was generated from real, between-
group differences.
For the comparisons of pre- and post-treatment, significant heterogeneity was
present in the domain, Shifting. However, the I2 was 72.41 suggesting at least 72%
of the variance was generated from real, between-time differences.
Moderator analysis, using a random effects model, was conducted to explore
potential, between-study differences that may explain this heterogeneity.
Moderator Analysis
Pre-treatment OSA to controls. Moderators investigated were age (Group 1
< 50, Group 2 ≥ 50 years), stringent inclusion/exclusion criteria (Group 1 = no,
Neurocognitive Disturbance in OSA 35
Group 2 = yes), control group selection criteria (Group 1 = PSG, Group 2 =
Questionnaire), and publication status (Group 1 = Published, Group 2 = Not
published). Papers were considered to have stringent inclusion/exclusion criteria if
they excluded participants with factors that could potentially affect cognition, such
as a history of traumatic brain injury, certain medications or diseases. None of these
moderators changed the effect significantly.
We were unable to examine the impact of disease severity as measured by AHI
as there were insufficient studies reporting effects of moderate OSA (AHI 15-29) in
each of the five sub-domains. A reviewer suggested dividing only those with AHI
≥30 into two groups and rerunning analyses. Accordingly, we divided the samples
into severe OSA (AHI 30 -50, N = 15) and very severe OSA (AHI 51+, N = 17).
Where there were sufficient samples (i.e. 2 or more) for each executive domain, all
effects remained significant and severity did not moderate the findings.
Pre-treatment to post-treatment differences. Between-study differences
were also investigated using moderator analysis. Moderators were age, stringent
inclusion/exclusion criteria, control selection criteria (PSG or questionnaire), length
of CPAP use, and publication status. No variables significantly moderated the effect
of CPAP treatment on the executive burden of OSA.
Likewise, disease severity, as measured by AHI, could not be examined as
there were insufficient studies reporting effects of moderate OSA in each of the five
sub-domains. As above, as recommended by a reviewer, we explored the impact of
severity within individuals with AHI ≥ 30. As before, all effects remained
significant, and severity did not moderate the findings.
Furthermore, we were unable to examine the impact of CPAP compliance on
effect sizes as only one study divided the participants into compliant and non-
Neurocognitive Disturbance in OSA 36
compliant users, all other studies excluded non-compliant individuals or reported
only the group mean compliance which was always above 4hrs.
However, and as recommended by a reviewer, we explored months on CPAP
by dividing samples into short (0-5 months, N = 13) and long (≥5.5 months, N = 5)
term CPAP use. Where there were sufficient samples (i.e. 2 or more) for each
executive domain, all effects remained significant and months on CPAP did not
moderate the findings.
Risk of Publication Bias
There is evidence to suggest that studies with a significant result are more
likely to be published, and that published studies are more likely to be available for
meta-analysis. (Borenstein et al., 2009) Publication bias was inspected visually using
funnel plots. These were asymmetrical, indicating the presence of bias. Egger’s test
for asymmetry (Egger, Smith, Schneider, & Minder, 1997) was used to investigate
this further.
Pre-treatment OSA to controls. For the OSA to control samples, Egger’s test
was non-significant for Inhibition (intercept 4.79; 95% CI: 0.04 to 9.54; p = 0.05),
and Updating (intercept 5.84; 95% CI: -0.63 to 12.32; p = 0.07), but was significant
for Fluid Reasoning (intercept 4.49; 95% CI: 0.84 to 8.13; p = 0.02), Generativity
(intercept 4.55; 95% CI: 0.65 to 8.45; p = 0.03) and Shifting (intercept 4.48; 95% CI:
1.72 to 7.25; p = 0.002). However, Rosenthal’s Fail-safe N, which represents the
number of studies needed to create an overall non-significant effect, (Rosenthal,
1979) was 172 non-significant studies for Fluid Reasoning, 49 studies for
Generativity and 232 studies for Shifting.
Duval and Tweedie’s Trim and Fill procedure (Duval & Tweedie, 2000) was
used to determine the best estimate of an unbiased, overall effect size for the OSA cf
Neurocognitive Disturbance in OSA 37
control samples for the domains of Fluid Reasoning, Generativity and Shifting. For
Fluid Reasoning, the overall effect size was reduced from a large effect of 0.80 to a
small effect of 0.26. For Generativity the overall effect size shifted from a medium
effect of 0.59 to a small effect of 0.45. For Shifting, the overall effect size was
reduced from a medium effect of 0.53 to a small effect of 0.37. This suggests
publication bias may be inflating the estimates in the domains of Fluid Reasoning,
Generativity and Shifting, but that there were still significant differences between
those with OSA and controls in these domains.
Pre-treatment to post-treatment differences. For the pre- to post-treatment
effects, Egger’s test was non-significant for all five domains; Shifting (intercept
1.62; 95% CI: -1.39 to 4.63; p = 0.26), Generativity (intercept 0.70; 95% CI: -1.16 to
2.56; p = 0.39), Fluid Reasoning (intercept 0.17; 95% CI: -1.43 to 1.76; p = 0.82),
Inhibition (intercept 8.61; 95% CI: -8.29 to 25.52; p = 0.27) and Updating (intercept
0.86; 95% CI:-12.73 to 14.45; p = 0.89). This indicates no publication bias in these
domains.
Discussion
The current paper builds on previous reviews by focusing on executive
function within five, theoretically-driven subcomponents. Three questions were
posed: 1) which specific executive functions are affected by the presence of
untreated OSA?; 2) if executive functions are impaired, does treatment help to
remediate these deficits?; and 3) are any of these effects moderated by publication
status, sample source, study design, age, disease severity, treatment length or control
screening?
Neurocognitive Disturbance in OSA 38
The findings of the present review
The results from the present analysis indicate that executive function is
impaired in OSA compared to control participants across all five subcomponents.
People with OSA have difficulty Shifting between tasks or mental sets, Updating and
monitoring working memory representations, Inhibiting dominant or pre-potent
responses, they struggle with Generating new information without external input or
efficiently accessing long term memory, and they have significant problems with
Fluid Reasoning or problem solving. Further to this, the present research
demonstrated that if participants undertake CPAP treatment, executive function
difficulties across these five sub-domains are reduced.
This meta-analysis was unable to assess the impact of CPAP compliance on
improvement in executive function. Articles assessed in the present review excluded
individuals who were not compliant with treatment, or reported only the group mean
number of hours CPAP was used, except in one instance where participants were
divided into compliant or non-compliant, hence we cannot make any concluding
statement on improvements in executive function in individuals who do not follow
their treatment regime. However, exploration of the impact of months of CPAP use
revealed no additional gain with extended use (6 months or more).
A recent review (Bucks et al., 2012) summarised the current understanding of
cognitive function across a number of domains including executive function. The
authors found that in 2 reviews (Aloia et al., 2004; Saunamäki & Jehkonen, 2007)
and 2 meta-analyses (Beebe et al., 2003; Fulda & Schulz, 2003) meeting inclusion
criteria, executive function was impaired by comparison with controls and norms.
The present results provide further support that executive function is impaired in
people with OSA. Furthermore this review (Bucks et al., 2012) examined
Neurocognitive Disturbance in OSA 39
improvement in executive function following CPAP treatment in 1 meta-analysis
(Aloia et al., 2004) and 1 literature review. (Saunamäki & Jehkonen, 2007) The
reviews examined came to opposing conclusions, hence the summary was
inconclusive. The present meta-analysis suggests that, overall, CPAP treatment is
successful in improving executive function difficulties caused by OSA, and adds to
the available evidence supporting the benefits of CPAP treatment.
Past reviews have grouped executive function into one combined domain, or
collapsed them by test, making it impossible to delineate the subcomponents of
executive function that are impaired. The present paper views executive dysfunction
within current neuropsychological understanding of executive function. Such a
framework provides a possible explanation for relationship or work difficulties seen
in OSA. Individuals with OSA may experience relationship (Engleman & Douglas,
2004; Kales et al., 1985) or work productivity difficulties (Hillman et al., 2006b;
Mulgrew et al., 2007) as they may not be able to inhibit inappropriate responses to
aggravating social situations, or solve novel problems in a work place with Fluid
Reasoning.
The effect of moderators
The present study was not able to examine the impact of disease severity on
OSA across the full range of AHI, as there were insufficient numbers of mild (AHI
5-14) and moderate samples (AHI 15-29). However, comparison of severe (AHI 30-
50) and very severe (AHI 50+) OSA samples revealed no impact of severity on the
deficits found, and no impact on executive consequences of CPAP treatment. The
literature is divided with regard to whether there is a relationship between disease
severity as measured by AHI and cognitive dysfunction. (Aloia et al., 2004; Carlson,
Neelon, Carlson, Hartman, & Bliwise, 2011; Tsai, 2010) In a recent meta-analysis of
Neurocognitive Disturbance in OSA 40
episodic memory function, Wallace and Bucks (Wallace & Bucks, 2012) found no
relationship with disease severity and in a systematic review Aloia et al. (Aloia et al.,
2004) found no relationship between disease severity and executive function
however they did find a positive relationship between disease severity and global
cognitive function and attention/vigilance. As yet, the link between OSA disease
severity and cognition is unclear. This is most likely due to the complex picture of
co-morbidity, and as yet, no definitive way of measuring disease onset in OSA.
(Valencia-Flores, Bliwise, Guilleminault, Cilveti, & Clerk, 1996)
Previous studies have demonstrated that age and OSA results in a double
burden, with older individuals exhibiting poorer cognition. (Antonelli-Incalzi et al.,
2004; Ayalon, Ancoli-Israel, & Drummond, 2010) The present meta-analysis did not
find this same relationship as age did not significantly moderate the results. In
studies comparing older and younger participants with OSA, older adults have been
found to be more impaired on tests of executive functioning. (Antonelli-Incalzi et al.,
2004; Ayalon et al., 2010) The lack of effect in the current meta-analytic review may
be due to assessing the effect of age using group averages, which resulted in similar
age distributions across samples. Primary comparison studies which explore the
interaction of age and OSA on these executive function subcomponents will be
important for clarifying this relationship.
Furthermore, selection of controls with PSG or questionnaires did not
significantly moderate the findings. This result may seem surprising given that
estimates of undiagnosed OSA are high. (Al-Lawati et al., 2009; Butkov & Lee-
Chiong, 2007; Young et al., 1997) However, this finding is consistent with that
recently reported by Wallace and Bucks. (Wallace & Bucks, 2012) Whilst it is still
the case that some control participants in primary studies may have undiagnosed
Neurocognitive Disturbance in OSA 41
OSA, including studies which have used questionnaire screening procedures does
not appear to confer a risk of failing to find an effect in meta-analyses in OSA.
Heterogeneity among results
Many of the domains demonstrated a high level of heterogeneity in the test
results, indicating that there may be other factors in each domain influencing the
observed mean. However, further analysis demonstrated this heterogeneity did not
obscure real differences in each subdomain of EF. The domain that exhibited the
most heterogeneity was the domain with the largest number of different tests, Fluid
Reasoning, in which there were 13 different tests. Neurocognitive tests, especially
executive function tests, even purportedly measuring the same domain or sub-
domain, will also capture facets of other cognitive or motor skills. For example
Trails B, a measure of Shifting, also requires attention and taps into psycho-motor
speed. (Lezak et al., 2004) Furthermore, few tests currently used to assess executive
function were originally designed for the specific purpose of measuring executive
function. (Banich, 2009; Suchy, 2009)
Future research exploring the executive dysfunction of OSA may benefit from
selecting measures more closely targeted at executive functions and designed to
fractionate performance into theoretically-driven and dissociable subcomponents.
One such measure is the Random Number Generation (RNG) task. (Maes, Eling,
Reelick, & Kessels, 2011) This task takes only a few minutes, is easily administered
with a laptop computer and provides measures of Shifting and Inhibition. (Maes et
al., 2011; Miyake et al., 2000)
One other factor that might lead to heterogeneity is pre-morbid IQ or
intelligence. This is because greater IQ and/or education appears to provide
‘protection’ against cognitive decline because of greater cognitive reserve.
Neurocognitive Disturbance in OSA 42
(Alchanatis et al., 2005; Stern, 2002) Unfortunately, not all studies provided a
measure of academic achievement or pre-morbid intelligence (IQ). Given that age
decreases reserve and IQ increases it, primary studies that stratify the sample into
age and IQ groups when examining the impact of OSA on executive function are
needed.
Limitations
An issue that cannot be addressed by the current review is whether the deficits
evidenced in the literature are primary or secondary. That is to say, whether the
deficits found in OSA in executive function are due to neurological damage (primary
effect) or to impairments in attention which themselves are the result of sleep
fragmentation, sleep deprivation and the associated excessive daytime sleepiness.
(Verstraeten & Cluydts, 2004; Verstraeten et al., 2004). Given evidence of frontal
activation in participants completing these tasks (Zhang et al., 2011) and the
presence of structural abnormalities in the frontal lobes of individuals with OSA
(Zimmerman & Aloia, 2006), it seems likely that these executive function deficits
are a primary and direct consequence of OSA (see Beebe and Gozal (Beebe &
Gozal, 2002) for a review; but see Durmer and Dinges (Durmer & Dinges, 2005b)
for evidence that similar frontal changes are also seen post sleep loss). Indeed, one
study, (Verstraeten et al., 2004) which controlled for attention deficits while
exploring executive dysfunction differences between OSA and controls, concluded
that most of the executive difficulties were secondary to attention problems. This
study demonstrated that the one area with deficits remaining after controlling for
attention, was Shifting. More studies of this nature and the development of tasks that
can tease apart the contributions of attention and executive cognitive processes to
task performance are required.
Neurocognitive Disturbance in OSA 43
Conclusions
People with OSA have difficulty with the executive facets of Shifting,
Updating, Inhibiting, Generativity, and with Fluid Reasoning. Further, the present
research indicates that all these difficulties improve with CPAP treatment. Age and
disease severity did not moderate the effects found, however, further studies are
needed exploring the extent of primary and secondary effects, and the impact of age
and pre-morbid ability (cognitive reserve).
Neurocognitive Disturbance in OSA 44
CHAPTER 3:
RECRUITMENT, PARTICIPANTS AND MATERIALS
This chapter presents a detailed description of recruitment processes,
participant details, and the materials used in Chapters 4 and 5 of this thesis, given
strict word limits of the journals in which the papers were published.
The present thesis collected two samples; one sample of individuals with
diagnosed sleep disorders from the West Australian Sleep Disorders Research
Institute, Sir Charles Gairdner Hospital (the clinical sample), and another from the
Western Australian Participant Pool, University of Western Australia (the
community sample). The studies presented in this thesis received approval from the
Human Research Ethics Committees of Sir Charles Gairdner Hospital and the
University of Western Australia. Data are used in Chapters 4 and 5 of this thesis.
Chapter 4 uses the questionnaire data from the clinical and community samples
(356 individuals from the community group, and 679 individuals from the clinical
group) to examine the factor structure of the Epworth Sleepiness Scale (the ESS).
Chapter 5 uses the sleep study data from the clinical and community samples
to examine the relationships between cognition, sleep fragmentation and hypoxia.
This chapter use the sample of participants who underwent polysomnography; 134
individuals from the clinical sample and 16 individuals from the community group.
Further details of these samples, recruitment processes and materials are
provided below.
Recruitment
Participants
Participants were 18 years of age or over, spoke English as a first language, did
not have any sleep disorders other than OSA, and gave written, informed consent.
Neurocognitive Disturbance in OSA 45
Community Sample (Control sample). Between January 2011 and
September 2011, 356 participants were recruited from the wider community through
community and University notices, community talks, radio announcements,
fundraising events, the University of Western Australian psychology student pool,
and the Western Australian Participant Pool (a panel of community volunteers aged
50, Director RS Bucks). Participants were aged 44±22 years (range 18 to 100 years),
BMI was 25.3±6.3 (range 15.5 to 62.1) and 248 (70%) were female. Data from these
participants were later used in Chapter 4 to examine the factor structure of the ESS.
From this wider sample, participants were selected through an algorithm
developed by Marshall, Dawson, and Bucks (2009) which identifies people at low
risk of sleep disordered breathing based on sleep disordered symptoms (e.g., snoring
and/or high BMI). A brief description of this algorithm is detailed below (See,
(Marshall et al., 2009). At the time the questionnaire was administered, participants
were also asked to indicate if they would agree to participate in an overnight sleep
study. Fifty participants who were low risk also indicated they would like to
participate in an overnight study. Of these 50, 24 attended an overnight sleep study at
the University of Western Australia, Centre for Sleep Research. The other 27
individuals declined when phoned, could not be contacted with the details provided,
or did not arrive on the night of their arranged study. The final sample was N = 16,
as used in Chapter 5 to examine the relationship between cognition and nocturnal
disturbance (Table 1). AHI in these individuals ranged from 0 to 16 breathing events
per hour. Typically, individuals assessed in a sleep clinic are ‘flagged’ for treatment
if they score an AHI of 15 or more. Thus, the community recruitment process
allowed us to recruit to the lower end of the OSA severity spectrum (from none to
Neurocognitive Disturbance in OSA 46
very mild OSA). Should hypoxia and sleep fragmentation affect cognition on a
continuum from no harm to great harm, these community participants provide the
data for the lower end of the spectrum.
Clinical Sample (OSA sample). Between March 2009 and July 2011, 679
participants were recruited consecutively through the West Australian Sleep
Disorders Research Institute, at Sir Charles Gairdner Hospital, as they came into the
sleep clinic for overnight assessment of a sleep disorder. Participants were aged
50.3±22.3 years (range 17 to 82 years), BMI was 33.0±7.4 kg.m2 (range 16.2 to
61.4), AHI was 34.8±29.7 events/hr (range 0 to 193) and 305 (44%) were female.
Data from these participants were later used in Chapter 4 to examine the factor
structure of the ESS. Demographic details are presented in Table 3.
Neurocognitive Disturbance in OSA 47
Table 3.
Demographic data for the clinical and community participants.
Community sample
N = 16
Clinical sample
N = 134
Whole sample
N = 150
M SD Min. Max. M SD Min. Max. M SD Min. Max.
Age 40.3 4.7 9 7 53.9 12.1 25 82 52.4 3.1 19 82
BMI 24.4 4.6 16.5 34.3 34.1 7.5 15.5 57.5 33.0 7.8 15.5 57.5
Gender 44% male 7 males 55% male 74 males 54% male 81 males
Education 12.7 2.9 10.0 20.0 11.6 3.0 6.0 23.0 11.9 3.1 6.0 23.0
Premorbid IQ 112.9 5.5 100.0 121.6 110.4 8.0 86.3 31.1 110.7 7.8 6.3 131.1
AHI 2.2 3.9 0 2.8 41.3 26.1 8.1 154.8 37.1 27.5 0.0 154.8
Note: Premorbid IQ as estimated using the NART; AHI = Apnoea Hypopnoea Index; BMI = Body Mass Index.
Neurocognitive Disturbance in OSA 48
Participants were individuals identified on overnight polysomnography
(PSG) as having clinically significant OSA, and who chose to complete the
questionnaires outlined below. Questionnaires were placed by the participants’
bedside by sleep technicians, and participants were invited to complete them, if
they so chose. Those who completed questionnaires were then invited to
complete neuropsychological evaluation at the University of Western Australia.
Participants were scheduled for neurocognitive assessment after receiving a
diagnosis of OSA, but before beginning CPAP treatment. These participants also
completed the set of neurocognitive assessments detailed below.
Of 296 participants approached for cognitive testing, 126 declined to
participate, 32 did not attend their appointment or cancelled, 2 discontinued testing
during the appointment, and 2 had begun CPAP treatment prior to their appointment.
Thus, the final sample of clinical (OSA) participants who completed pre-treatment
assessment was N = 134: for demographic details, see Table 3. Data from these
participants were used in Chapter 5 to examine the relationship between cognition
and nocturnal disturbance.
Materials
Questionnaires
Demographic questionnaire.
Body Mass Index (BMI). Participant BMI (Weight kilograms / Height in
metres2) was calculated based on medical records from the hospital, or taken when
the participant visited UWA for their overnight sleep study. If the participant was
only involved in the questionnaire portion of the study self-report data was used.
Epworth Sleepiness Scale (ESS) ((Johns, 1991). The ESS is an eight-item,
self-administered, subjective scale in which respondents rate their perceived
Neurocognitive Disturbance in OSA 49
likelihood of ‘dozing’ in a variety of everyday situations, e.g. ‘list an item here’.
Participants respond from 0 - ‘would never doze’ to 3 - ‘high chance of dozing’, with
an ESS score >10 representing excessive sleepiness (Johns & Hocking, 1997).
Strong internal consistency has been reported (Cronbach’s alpha 0.74 – 0.88; (Izci et
al., 2008; Johns, 1994; Smith et al., 2008), as well as good test-retest reliability ≥
.81(Izci et al., 2008). Use of the ESS is widespread in both clinical and research
settings (Bloch, Schoch, Zhang, & Russi, 1999; Chen et al., 2002; Izci et al., 2008;
Shahid, Shen, & Shapiro, 2010).
Depression, Anxiety, and Stress Scale-short form (DASS-21)(Lovibond &
Lovibond, 1995). The DASS-21 is a 21 item, self-administered questionnaire
designed to measure depressive, anxiety, and stress symptoms. The 21 item scale
was developed from the highest loading items on all 3 negative affect subscales
(Lovibond & Lovibond, 1995). Respondents indicate the degree to which they felt
negative affect symptoms over the past week, from 0 – ‘did not apply to me at all’ to
3 – ‘applied to me very much, or most of the time’. Scores are doubled to be
consistent with the original 42-item DASS and range from 0 to 126, with higher
scores indicating greater depression, anxiety and stress, respectively. The scale has
strong internal consistency on all three subscales (Depression 0.82 – 0.84; Anxiety
0.74 – 0.80; Stress 0.84 – 0.88; (Norton, 2007), good test-retest reliability .90 – .99
(Norton, 2007), and a reliable factor structure in clinical and non clinical samples
(Antony, Bieling, Cox, Enns, & Swinson, 1998; Henry & Crawford, 2005; Szabó,
2010).
Insomnia severity scale (ISI) (Bastien, Vallieres, & Morin, 1999). The ISI is a
7-item scale evaluating an individual’s satisfaction with their sleep patterns, daily
functioning due to current sleep habits and level of distress caused by their sleep.
Neurocognitive Disturbance in OSA 50
The items are rated from 0 – ‘not at all’ to 4 – ‘extremely’. Scores range from 0
through 28, with higher scores indicating greater dissatisfaction with current sleep
patterns. The scale has good internal consistency as measured by Cronbach’s alpha,
0.76 - 0.78 (Bastien et al., 1999), and adequate concurrent validity with other
assessments of non-restorative sleep, .65 (sleep diaries, wake after sleep onset)
(Morin & Azrin, 1985).
Pittsburgh sleep quality index (PSQI) (Buysse, Reynolds, & Monk, 1989).
The PSQI is a 19-item, self-rated questionnaire assessing subjective sleep quality
over the preceding month. These 19 questions are subdivided into 7, equally
weighted subscales, each assessing a different facet of sleep quality: Sleep duration,
Sleep disturbance, Sleep latency, Days of dysfunction due to sleepiness, Sleep
efficiency, Overall sleep quality, Need of medication to sleep. These subscale scores
are summed for a global PSQI score between 0 and 21, with higher scores indicating
poorer sleep quality. The scale has good internal consistency as measured by
Cronbach’s alpha, 0.72 - 0.83 (Buysse et al., 1989; Suleiman & Yates, 2011), and
excellent test-retest reliability, 0.85 (Buysse et al., 1989).
Berlin Sleep Questionnaire (Berlin) (Netzer, Stoohs, Netzer, Clark, & Strohl,
1999). The Berlin is a 10-item, self-rated questionnaire assessing an individual’s risk
of sleep-disordered breathing. The questions load onto 3 categories (snoring, fatigue,
obesity and hypertension). An individual is considered at high risk of Sleep
Disordered Breathing (SDB) if they score positively on 2 or more categories. SDB is
a meta-category for all nocturnal breathing disorders including OSA. The scale has
good internal consistency as measured by Cronbach’s alpha, 0.86 - 0.92 (Netzer et
al., 1999).
Neurocognitive Disturbance in OSA 51
Functional Outcomes of Sleep Questionnaire (FOSQ-10) (Weaver et al.,
1997). The FOSQ-10 is a 10-item, self-report scale evaluating functional outcomes
as a result of sleep quality, developed from the highest loading items for each of the
5 subscales from the original 30-item questionnaire. Items are rated from 1 – ‘no’ to
4 – ‘yes, extreme’ and for questions 3, 4, 5 & 7, an option for 0 – ‘Don’t do this
activity for other reasons’. These items are divided into 5 subscales and a total score.
The mean of the subscale scores is multiplied by 5 to give a score of between 5 and
20, with higher scores indicating poorer functional outcomes. The scale has good
internal consistency, 0.86 – 0.91 (Weaver et al., 1997), and excellent test-retest
reliability, 0.81 – 0.90 (Weaver et al., 1997).
Questionnaire Data Processing.
Algorithm. The Algorithm selected for use in this study has been used to
identify high and low risk of sleep disordered breathing in a truck driving population
(Marshall et al., 2009). The algorithm uses the scores from the Berlin, ESS, PSQI
and FOSQ-10 questionnaires outlined above. This combination of questionnaires has
greater sensitivity (78.1) and specifity (93.5) than any of the individual
questionnaires alone (Marshall et al., 2009). For further detail about scoring this
algorithm, permission must be sought from the creators (Marshall et al., 2009)
The present study modified the algorithm to exclude participants if they were
at high risk of restless legs syndrome (answered yes on the PSQI to ‘Have you had
trouble sleeping due to your legs twitching or jerking while you sleep?), or those
with high levels of depression on the DASS (scores of 20 or more), or those with
high scores on the ISI (scores of 10 or more). This was done to exclude participants
who were at high risk of comorbid sleep disorders (restless legs syndrome and
insomnia), and individuals with high levels of depression symptoms. These disorders
Neurocognitive Disturbance in OSA 52
carry their own cognitive dysfunction profile that may have interfered with the
investigation of the cognitive profile of OSA (Burt, Zembar, & Niederehe, 1995;
Fulda & Schulz, 2001; Pearson et al., 2006). Neurocognitive Assessment
Participants were assessed by a graduate psychology student trained to
administer the tasks. Assessments were scored twice, the first time immediately by
the assessor, and a second time by another trained assessor. In the event there were
any scoring discrepancies, these issues were discussed with a third person (RSB).
Assessors followed a scripted protocol that outlined when and how tasks were to be
administered, and what instructions were to be given. Timed tasks were assessed
using a stopwatch. Participants completed the following neurocognitive assessments:
Cognitive Drug Research System (CDR) (Corporation, 2009; Wesnes,
2000). The CDR is a 30 minute, computerized battery of cognitive assessments
measuring attention, short-term and episodic long-term memory. The tasks were
administered on a 15-inch screen laptop computer. Word tasks were all presented in
white, capitalized print on a dark blue background. All tasks have alternate versions,
which are presented in random order. Alternate versions of the tasks were presented
to OSA participants who returned for follow up testing (after 3 months of CPAP
treatment) to control for practice effects. Participants respond verbally or using a
two-button response box, marked YES and NO.
Participants were instructed to answer as quickly and accurately as possible
during each of the tasks. These tests have good reliability and validity (van den Goor
et al., 2008; UBC, 2010). The CDR has previously been used to assess subtle
cognitive changes in OSA (Hirshkowitz et al., 2007; Roth, Rippon, & Arora, 2008;
Roth et al., 2006) and other chronic illnesses, such as depression and mild cognitive
impairment (Newhouse et al., 2012; Vasudev et al., 2012), with good reliability and
Neurocognitive Disturbance in OSA 53
validity (Van Den Goor et al., 2008), and has been described in detail elsewhere
(Wesnes, 2000).
Assessment of Attention.
CDR Digit vigilance. A single digit number, the target number, between zero
and nine (inclusive) was presented to the right of the screen. A series of single digit
numbers was also presented in the centre of the screen, and scrolled though
randomly at a rate of 150 numbers per minute. The target number was presented a
total of 45 times at random intervals. This task yielded a measure of the % of targets
detected accurately (min = 0, and max = 100).
CDR Choice reaction time. The word ‘YES’ or ‘NO’ was randomly presented
on screen, for a total of 30 presentations, with an inter-stimulus interval ranging from
one to three and a half seconds. The participant used the response box to respond
using the ‘yes’ button every time they saw ‘YES’ or the ‘no’ button if they saw ‘NO’
as quickly as possible. This task yielded a measure of the % of targets detected
accurately (min = 0, and max = 100).
Assessment of Long Term Memory.
CDR Picture presentation and delayed recognition. Twenty full colour, high
resolution images were presented on screen, one at a time for one second each with a
three second space between images. Participants were instructed to remember the
image as later they would be asked to recall the images. Twenty minutes later, the
original twenty pictures were presented one at a time, in random order, amongst
twenty distracter pictures. The participant was instructed to respond using the
response box by pressing ‘yes’ if they had seen the picture before or ‘no’ if the
picture was new. This measure gave ‘% greater than chance’ accuracy (% original
Neurocognitive Disturbance in OSA 54
targets correctly identified + % novel targets correctly identified – 100; min = 0, and
max = 100).
CDR Delayed word learning. Fifteen words are presented for one second in
the centre of the monitor, with a two second space between words. The participant is
instructed to remember as many of these 15 words as they are able, as they will be
asked to recall them at a later stage. Both immediately post learning, and after a
delay, the participant was asked to recall as many of the words from the original list
of 15 words as possible. This task gave a measure of the amount of new information
the participant had recalled (T1_IRCL - T1_DRCL; min = -15, and a max = 15).
CDR Delayed word recognition. The 15 words from the original word
presentation task were presented one at a time, in random order, amongst fifteen
distracter words. The participant was instructed to respond using the response box by
pressing ‘yes’ if they had seen the word before or ‘no’ if the word was new, for each
of the 30 words as quickly as possible. This task yielded a measure of the ‘% greater
than chance’ performance (% original targets correctly identified + % novel targets
correctly identified – 100; min = 0, and a max = 100).
Assessment of Short Term Memory.
Short term numeric memory. The participant was instructed to remember five
single digits between zero to nine (inclusive) presented in series in the centre of the
screen at the rate of one digit per second. The instructor checked that the participant
recalled the numbers correctly by asking them to recite the numbers aloud. A second
presentation of the numbers was administered in the event the participant could not
recite the numbers correctly. Following this, a series of single digit numbers was
presented, to which the participant responded ‘yes’ if it was one of the original they
were remembering and ‘no’ if it was not one of the original five. This task yielded a
Neurocognitive Disturbance in OSA 55
measure of ‘% greater than chance performance’ (% original targets correctly
identified + % novel targets correctly identified – 100; (min = 0, and a max = 100).
CDR Spatial working memory. The participant was shown a laminated card
with a pictorial representation of a house with nine windows and instructed that the
same image would be presented on screen with three windows lit. A different
configuration of ‘lit’ and ‘dark’ windows was presented a total of 36 times. The
participant was instructed to remember the placement of the lit windows for the
activity to follow. Then an image of the same house with one lit window was
presented. Using the response box the participant was instructed to respond ‘yes’ if
the lit window was one of the original 3 lit windows, or ‘no’ if the window was dark
in the original house. This task yielded a measure of ‘% greater than chance
performance’ score (% original targets correctly identified + % novel targets
correctly identified – 100; min = 0, and a max = 100).
Assessment of Executive Function.
Clock Drawing Task 1 & 2 (CLOX 1 & 2) (Royall, Cordes, & Polk, 1998).
CLOX is a brief, clock drawing task that screens for visuospatial/construction and
executive function difficulties, specifically the strategic control and planning of
behaviour (Strauss et al., 2006). CLOX scores correlate highly with other executive
function tasks (Strauss et al., 2006). Participants draw a clock without assistance
(CLOX 1) and then copy the examiner’s clock (CLOX 2). An executive control
function score (ECF) is calculated by subtracting the CLOX 1 score from the CLOX
2 score. Higher ECF scores indicate better executive function. The scale has
excellent internal consistency, 0.82 (Royall et al., 1998), and inter-rater reliability,
0.93 (Royall et al., 1998).
Neurocognitive Disturbance in OSA 56
FAS Controlled Oral Word Association task (COWA) (Lezak et al., 2004).
The COWA is a 3-minute phonemic fluency task (Strauss et al., 2006), that taps into
executive function, specifically efficiency of access to long term memory stores
(Fisk & Sharp, 2004). The FAS has high internal consistency, 0.83 (Tombaugh,
Kozk, & Rees, 1999). Participants generate as many words as possible in 1 minute,
beginning with a given letter (F, A and S). The total repetitions, non-words and
intrusions are subtracted from the total number of words recalled to give the final
score. Higher scores indicate better performance.
Trail Making Test (TMT; Trails A and B) (Partington & Leiter, 1949; Reitan,
1958). The Trail Making Test is a five minute assessment in which the participant
connects circles numbered from 1 to 25 (Part A), or alternates between numbers and
letters in order (1-A-2-B, etc: Part B). The task measures attention, speed, and
mental flexibility. Test-retest reliability is good to excellent (Part A, 0.55 - 0.79; Part
B, 0.75 - 0.89), and inter-rater reliability is, likewise, excellent 0.89-0.92(Strauss et
al., 2006). Construct validity is good: trails correlates well with other measures of
attention, and part B is related to other measures of executive function (e.g. WCST
(Kortte, Horner, & Windham, 2002)). Longer times and more errors indicate poorer
performance. A ratio score (B/A), indexing the relative increase in time taken to
complete Part B as a function of Part A, was calculated (Lamberty, Putnam, Chatel,
Bieliauskas, & Adams, 1994) and used as a measure of set shifting/mental flexibility.
Assessment of Pre-morbid Intelligence.
National Adult Reading Test - 2 (NART-2) (Nelson & Willison, 1991). The
NART-2 is a 5-10 minute reading task providing an estimate of pre-morbid IQ.
NART scores are highly correlated with IQ in healthy participants (Schretlen,
Buffington, Meyer, & Pearlson, 2005) and with premorbid IQ and general cognitive
Neurocognitive Disturbance in OSA 57
ability (Crawford, Deary, Starr, & Whalley, 2001; Crawford, Stewart, Cochrane,
Parker, & Besson, 1989). Participants read aloud 50 irregular English nouns. Error
scores were combined with demographic variables to estimate pre-morbid
intelligence (Crawford et al., 2001; Crawford et al., 1989). The NART is robust to
declines in other cognitive skills, hence it has utility as a measure of premorbid
intelligence (Crawford et al., 2001).
Predicted WAIS FSIQ = 135.96 - 0.789 (NART errors) - 4.6 (sex) - 2.15 (social
class) + 0.122 (age)
For social class, 1 = professional, 2 = intermediate, 3 = skilled, 4 = semi-
skilled, 5 = unskilled. People who were currently retired, unemployed were coded by
their main lifetime occupation. Participants who had never worked were coded as
unskilled, traditionally wives were coded by their husbands occupation (Crawford et
al., 2001).
Sleep Metrics
Polysomnography (PSG)
PSG is the gold standard method of OSA diagnosis (Kryger, 2010).
Measurements of brain activity (sleep state and arousal) (EEG), eye movements
(EOG), skeletal muscle activation (EMG), heart rhythm (ECG), air flow (thermistor
and nasal pressure), breathing related effort (inductance plethysmography), blood
oxygen levels (pulse oximetry), snoring (vibration sensor), and patient position
(position monitor) are recorded. These measurements allow calculation of a variety
of measures of disease severity including level of sleep fragmentation and oxygen
desaturation. Sleep metrics were recorded using the Compumedics Grael HD-PSG
Sleep Diagnostic Amplifier System at the Centre for Sleep Research, University of
Western Australia and the Compumedics E-Series PSG/EEG at WASDRI, Sir
Neurocognitive Disturbance in OSA 58
Charles Gairdner Hospital. All sleep studies were scored using the Compumedics
PSG 3 software.
For the clinical participants, the sleep study was performed at the West
Australian Sleep Disorders Research Institute (WASDRI) at Sir Charles Gairdner
Hospital.
Sleep study scoring and staging
All sleep studies, whether recorded at WASDRI or UWA, were scored by the
researcher (MO) from the original, overnight polysomnography recordings. All
equipment placement, sleep staging and event scoring was completed according to
The AASM Manual for the Scoring of Sleep and Associated Events: Rules,
Terminology and Technical Specifications (AASM, 2012). A combination of EEG,
EOG and EMG are used for staging sleep, whilst the other sensors described below
are used for detecting and grading the severity of sleep disrupting
abnormalities/events.
Equipment Placement.
Table 4 provides a list of the measurements obtained. These measurements
were used for marking key events that may disturb the participants’ sleep and for
staging sleep.
Neurocognitive Disturbance in OSA 59
Table 4.
Devices used and measurements obtained during overnight polysomnograph.
Number of
electrodes
Placement Measures Purpose
Electroencephalogram
(EEG)
Six exploring
and three
reference
electrodes
Scalp in
accordance with
international
10-20 system
Brain activity Sleep staging
Electro-oculogram
(EOG)
Two exploring
and one
reference
electrode(s)
Outer canthus
of the left and
right eyes
Eye
movements
Sleep staging
(especially
REM sleep)
Electromyogram
(EMG)
Two exploring
and one
reference
Chin and
anterior tibialis
of the left and
right leg
Muscle tension Sleep staging
(especially
REM sleep) or
diagnosing
muscle
disorders
Electrocardiogram
(ECG)
Two exploring
electrodes
5th intercostal
space below the
left nipple and
right clavicle
Heart rate and
rhythm
Heart rate and
rhythm
Pulse Oximetry One electrode A finger Blood oxygen
saturation
Blood oxygen
saturation
Nasal Pressure One nasal
cannula
Inside the
nostrils
Nasal pressure Nasal airflow
Thermistor One
thermocouple
The base of the
nose and over
the mouth
Air
temperature
Nasal and oral
airflow
Neurocognitive Disturbance in OSA 60
Respiratory Inductance
Plethysmography
Two exploring
electrodes
Two bands –
one across the
abdomen and
one across the
thorax
Movement of
chest and
abdomen walls
Breathing
rhythm and
effort
Microphone One electrode Placed over the
bed or on the
throat
Sound To capture
snoring
Video Video camera Above the bed Video Capture gross
body
movements
Electroencephalogram electrode placement (EEG). EEG electrodes were
placed according to the international 10-20 system at the recommended placements;
Frontal 4 (F4), Central 4 (C4) and Occipital 2 (O2), referenced to Mastoid 1 (M1).
Backup electrodes were placed over Frontal 3 (F3), Central 3 (C3) and Occipital 1
(O1), referenced to Mastoid 2 (M2). Figure 4 shows the placement of the EEG. The
EEG records brain activity and is used for staging the presence or absence and the
depth of sleep.
Electrooculogram (EOG). EOG were placed according to the recommended
settings at Eye 1 (E1) referenced to M2 and Eye 2 (E2) referenced to M1. These
were placed 1 cm out and 1 cm up of the outer canthus of the right eye, and 1cm out
and 1cm down of the outer canthus for the left eye. The EOG records eye
movements.
Electromyogram (EMG). Three electrodes were used to record chin EMG; 1
in the midline 1 cm above the inferior ridge of the mandible, and 2 placed 2cm
Neurocognitive Disturbance in OSA 61
below the inferior ridge of the mandible, 1 was placed 2 cm to the right of the
midline and 1 placed 2cm to the left of the midline. Four electrodes were used for
measuring leg EMG. Two were placed on the body of each anterior tibialis, between
2 and 5 cm apart. Both Chin and Leg EMG record muscle activity in the region they
are applied.
Electrocardiogram electrode placement (ECG). ECG electrodes were
placed according to the recommended II lead torso electrode placement. One
electrode was placed on the middle of the left clavicle, the other electrode was
placed at intercostal space IV directly below the nipple. The ECG recorded the rate
and regularity of the person’s heartbeat.
Snore sensor/microphone. A snore microphone was placed on the
individual’s neck to the side of the participant’s laryngeal prominence at which the
Grael headbox sat (i.e. if the headbox was to the left of the bed, the sensor was
placed to the left of the laryngeal prominence). This placement was for patient
comfort. The snore microphone recorded the presence or absence of snoring and
other vocalizations.
Peizoelectric abdomen and thoracic bands. Two peizelectric bands were
used for monitoring breathing rate and rhythm. One band was placed around the
abdomen at the navel, and another around the thorax at or below the nipples and
under the arms. The bands are used for detecting effort during breathing. Figure 8
shows the placement of the piezoelectric bands.
Thermistor. The thermistor is placed at the base of the nose, with a sensor
over the mouth and in each nostril. The sensor records the rate and rhythm of
breathing through temperature. The thermistor is used for detection of apnoea.
Neurocognitive Disturbance in OSA 62
Nasal cannula. The nasal cannula is placed at the base of the nose with a
sensor in each nostril. The sensor records the rate and rhythm of breathing through
nasal pressure. The nasal cannula is used for detection of hypopnoea.
Oximeter. The oximeter was placed on a finger on the hand closest to the
Grael headbox. This placement was for patient comfort. The oximeter samples the
percentage of blood oxygen every 3 seconds.
Position sensor. The position sensor was placed around the thoracic
piezoelectric band at the centre of the person’s chest. This sensor registered the
position of the sleeping participant either as supine, prone or lateral right or left.
Video and audio. Video and audio were recorded with an infrared video
camera in the ceiling of each room. This captured the individual’s activity and
assisted with the definition of sleep stages and events when signals were ambiguous.
Sleep stage scoring (AASM, 2012)
Sleep was scored in 30-second, sequential epochs from the start of the study to
the end of the study. In the event that there were 2 or more stages in 1 epoch, the
epoch was assigned the sleep staging comprising the greater portion of the epoch.
Wake (W). Stage wake was scored if there were trains of alpha rhythm (8-13
Hz) over O1 and/or O2 when the participants eyes were closed, that attenuated with
eye opening; eyeblinks were present at a frequency of 0.5-2 Hz; reading eye
movements were evident, or if irregular eye movements were present with high chin
EMG.
Non-REM Stage 1(N1). N1 was scored if the reasonably regular slow eye
movements were present for more than 500 msec; if a low amplitude mixed
frequency (4-7Hz) EEG activity was present; if vertex sharp waves were present
Neurocognitive Disturbance in OSA 63
over C3 or C4, for less than .5sec. Sleep onset was scored as the first epoch with
greater than 50% sleep, this stage is usually N1.
Non-REM Stage 2 (N2). N2 was scored if K complexes unassociated with
arousals, or sleep spindles were evident and then continuously scored until the
presence of an arousal, major body movement or the person transitioned to N3 or
REM sleep. K-complexes are sharp negative waves immediately followed by a
positive component for between 0.5 to 1 second. A sleep spindle is a train of distinct
waves between 11-16Hz for longer than .5seconds.
Non-REM Stage 3 (N3). N3 was scored when the epoch exhibited greater
than 20% slow wave activity. Slow wave activity are waves of 0.5 to 2 Hz with peak
to peak amplitude of greater than 75 micro-volts.
Rapid Eye Movement Sleep (REM). REM was scored when low amplitude,
mixed frequency EEG; low chin EMG; saw tooth waves and/or rapid eye movements
were present. Rapid eye movements are irregular sharply peaked movements lasting
for less than 500msec. Saw tooth waves are serrated waves of 2 to 6 Hz often
preceding a burst of rapid eye movement.
Table 5 provides a summary of the key markers of sleep stages as defined by
the AASM manual (AASM, 2012).
Neurocognitive Disturbance in OSA 64
Table 5.
The neurological and physiological attributes of the different sleep stages in
healthy individuals (AASM, 2012).
Brain activity Muscle
activity
Eye activity Breathing
activity
Heart activity Other features
Wake Alpha brain
waves (8-13
Hz)
High
muscle tone
Reading eye
movements
Irregular,
sighs, yawns
Irregular Aware,
talking and
moving
Stage 1 (N1) Theta brain
waves (4-7
Hz)
Some loss of
muscle tone.
May exhibit
muscle
twitches
(myoclonus)
Slow rolling
eye
movements
Breathing
becomes
quite regular
Heart beat
becomes
quite regular
Shallow sleep
stage, can
jerk awake
readily
Stage 2 (N2) Theta waves
with sleep
spindles (11-
16 Hz) and
K-complexes
Decrease in
muscle tone
No eye
movements
Breathing
becomes
quite regular
Heart beat
becomes
quite regular
Decrease in
environmenta
l awareness
Stage 3 (N3) Delta waves
(0.5 – 2 Hz,
>75 μV
amplitude)
Decrease in
muscle tone
No eye
movements
Breathing
becomes very
regulated
Heart beat
becomes very
regulated
Deep or slow
wave sleep
REM Rapid low
voltage EEG
Loss of all
muscle tone
Rapid eye
movements
Breathing
becomes
quite irregular
Heart rate
becomes
quite irregular
Dreaming
sleep
Neurocognitive Disturbance in OSA 65
Event scoring (AASM, 2012)
Event scoring was completed in 5-minute epochs from the start of the study to
the end of the study.
Respiratory events. An apnoea was scored when there was a drop in the peak
thermistor sensor equal to or more than 90% for 10 seconds or more for at least 90%
of the event. The event is classified as obstructive if it is associated with increased
respiratory effort measured throughout the event. The apnoea is classified as central
if there is no inspiratory effort during the event. The event is classified as a mixed
apnoea if it demonstrates a lack of inspiratory effort during the first portion of the
event and effort is resumed in the last half of the event. A hypopnoea was scored
according to the alternative criteria, which state that a drop in nasal pressure equal to
or greater than 50% of baseline, lasting equal to or greater than 10 seconds for equal
or more than 90% of the event, accompanied with a 3% drop in oxygen saturation
from baseline (AASM, 2012).
Arousals. Arousals can be scored through any stage of sleep if there is an
abrupt shift in EEG frequency greater than 16Hz for more than 3 seconds, with a
minimum of 10 seconds of stable sleep preceding the arousal. During REM sleep,
this disturbance in the EEG must be accompanied by an increase in chin EMG.
Periodic Leg Movements (PLM). PLM was scored if the leg movement
lasted between .5 to 10 seconds with a greater than 8 micro-volt increase in baseline
leg EMG amplitude. There must be 4 such events in a row with between 5 and 90
seconds between each movement for them to be included in the same series.
Bruxism. Bruxism or tooth grinding was scored if the individual demonstrated
brief or sustained periods of elevated chin EMG of at least twice the baseline chin
Neurocognitive Disturbance in OSA 66
EMG for between .25 to 2 seconds in duration, for a minimum of 3 elevations.
Audio was also used to listen for the presence of tooth grinding sounds.
Measures of Disease Severity
Apnoea Hypopnoea Index (AHI). OSA severity is conventionally scored
using the AHI which is the average number of times a person exhibits an apnoea or
hypopnoea per hour of sleep (Kryger, 2010). According to published cut-offs, an
AHI of 0-4.9 is classified as ‘No OSA’, 5-14.9 as ‘Mild OSA’, 15-29.9 as ‘Moderate
OSA’, and 30+ as ‘Severe OSA’. People diagnosed and treated with mild OSA must,
additionally, have daytime complications not better explained by another illness (e.g.
depression, poor quality of life, high blood pressure) (Kryger, 2010).
Arousal Index (ArI). Sleep Fragmentation was operationalised using the
Arousal Index (ArI). The ArI is the total number of arousals divided by total sleep
time.
Oxygen Desaturation Index ≤3% of baseline (ODI3). ODI3 is the total
number of times the individual’s blood oxygen saturation dropped 3% below their
baseline oxygen saturation level.
Sleep Efficiency (SE). SE is the number of minutes of sleep divided by the
number of minutes in bed. Good SE is approximately 85 to 90%.
Mean and Lowest Sp02. Oxygen saturation is defined as the ratio of
oxyhemoglobin to the total concentration of hemoglobin present in the blood (ie
Oxyhemoglobin + reduced hemoglobin). When arterial oxyhemoglobin saturation is
measured non-invasively by pulse oximetry, it is called SpO2. Mean SpO2 is the
average SpO2. Lowest SpO2 is the lowest point SpO2 reached during sleep.
Neurocognitive Disturbance in OSA 67
Conclusions
The present chapter detailed the participants, recruitment, and materials used
in chapters 4 and 5 of the present thesis. These methods were used in part or full for
both samples; the clinical and community samples.
Neurocognitive Disturbance in OSA 68
CHAPTER 4: EXAMINING THE UTILITY OF THE ESS
Preface
Excessive daytime sleepiness (EDS) is both a common symptom of OSA and
impacts on cognitive function. EDS can be measured subjectively with
questionnaires such as the Epworth Sleepiness Scale (ESS), or objectively with tasks
such as the Multiple Sleep Latency Test (MSLT). There is debate as to which
method is best for ascertaining daytime sleepiness. Considered in this debate is that
the ESS far outweighs the MSLT for time and monetary cost, and may be more
ecologically valid, as it is a simple paper-and-pencil task taking approximately five
minutes to complete, and asks about sleepiness in a range of circumstances most
people are exposed to daily. Possibly due to ease and low cost of administration and
high ecological validity, the ESS is used extensively (Shahid et al., 2010).
The present thesis uses the ESS as a measure of sleepiness in order to
understand the impact of sleepiness on cognition. Prior to doing so, the authors
evaluated the factor structure of the ESS as a measure of sleepiness in both a
community and clinical sample.
This chapter was published in the journal Sleep and Breathing: Olaithe M.,
Skinner T.C., Clarke J., Eastwood P., Bucks R.S. (2012). Can we get more from
the Epworth Sleepiness Scale than just a single score? A confirmatory factor
analysis of the ESS. Sleep and Breathing; 17(2); 763-9.
Neurocognitive Disturbance in OSA 69
Abstract
Purpose: The Epworth Sleepiness Scale (ESS) is a widely used tool for measuring
sleepiness. In additional to providing a single measure of sleepiness (a one factor
structure), the ESS also has the capacity to provide additional information about
specific factors that facilitate sleep-onset, including a persons posture, activity and
environment. These features of sleepiness are referred to as somnificity. This study
evaluates and compares the fit of a 1-factor structure (sleepiness) and 3-factor structure
(reflecting low, medium and high levels of somnificity) for the ESS.
Methods: All participants (a community sample N = 356 and a clinical sample N =
679) were administered the ESS. Confirmatory Factor Analysis was used to evaluate
and compare the fit of 1-factor and 3- factor models of the ESS.
Results: In both samples, a 3-factor structure (community sample adjusted X2 = 2.95,
RMSEA = 0.07, CFI = 0.95; clinical sample adjusted X2 = 3.98, RMSEA = 0.07, CFI
= .98) provided a level of model fit that was at least as good as the 1-factor structure
(community sample adjusted X2 = 5.01, RMSEA = 0.11, CFI = 0.87; clinical sample
adjusted X2 = 8.87, RMSEA = 0.11, CFI = 0.92).
Conclusions: In addition to a single measure of sleepiness, the ESS can provide
subscale scores which relate to three underlying levels of somnificity. These findings
suggest that the ESS can be used to measure an individual’s overall sleep propensity
as well as more specific measures of sleep propensity in low, moderate and high levels
of situational somnificity.
Neurocognitive Disturbance in OSA 70
Can we get more from the Epworth Sleepiness Scale than just a single score? A
confirmatory factor analysis of the ESS
Excessive Daytime Sleepiness (EDS) is a subjective report of a lack of
energy during the day and an amplified desire to fall asleep when sedentary, despite
a full nights’ sleep (Sauter et al., 2010; Tsai, 2010). EDS is associated with an
increased risk of motor vehicle accidents (Lyznicki, Doege, Davis, & Williams,
1998; Webb, 1995), decreased quality of life (Takegami et al., 2011) and significant
financial costs within the community (Webb, 1995).
The Epworth Sleepiness Scale (ESS) (Johns, 1991, 2002) is a simple,
commonly used measure of sleepiness among a variety of populations in both
clinical and research settings (Bloch et al., 1999; Chen et al., 2002; Izci et al., 2008;
Shahid et al., 2010). Its widespread use has resulted in the ESS becoming highly
influential in shaping our understanding of the concept of sleepiness and its response
to treatment (Johns, 2002). Developed by Johns in 1991 (Johns, 1991) the ESS
generates a single score from estimates of the level of sleepiness an individual
experiences in eight commonly encountered situations, each question reflecting a
low, medium or high somniforic (sleep-inducing) situation (Johns, 1994). While
generally used as a single score, previous research demonstrates that the ESS has the
potential to provide additional information about an individual’s sleepiness by
analysing it in terms of 3 factors, reflecting low, medium and high somniforic
situations (Johns, 1994).
Confirmatory factor analysis (CFA) is a flexible statistical technique
designed for construct validation and scale refinement (MacCallum & Austin, 2010).
Neurocognitive Disturbance in OSA 71
In the present study we have applied it to examine the hypothesised relationships
between the construct (sleepiness/somnificity), and the items used to measure this
construct (the 8 ESS items) (MacCallum & Austin, 2010). CFA is particularly useful
when examining hypothesised constructs such as sleepiness as it can account for
measurement error that naturally exists between the ‘pure’ construct and the
measurement of the construct (Iaccobucci, 2010), providing a stringent test of this
latent structure. Because CFA allows the comparison of alternative models, and
identifies the model of ‘best fit’ to the data (Byrne, 2010; Floyd & Widaman, 1995),
it is an ideal technique for assessing the fit of different models of the ESS.
A number of studies have investigated the factor structure of the ESS (Chen
et al., 2002; Heaton & Anderson, 2007; Izci et al., 2008; Johns, 1992, 1994;
Kingshott, Engleman, Deary, & Douglas, 1998; Smith et al., 2008), however none
has compared the original 1-factor design to a 3-factor structure. Therefore, the aim
of the current study was to use CFA to compare a 1-factor to a 3- factor structure
(reflecting low, medium and high somnificity) in community and clinical participant
samples. Study of these populations is important as the ESS is commonly used in
both clinical and community settings.
Method
The ESS was completed by a community sample and a clinical sample, either
online or in hardcopy, as part of a set of questionnaires investigating their subjective
sleep, mood and, memory functioning. The study was approved by the Human
Research Ethics Committees of Sir Charles Gairdner Hospital, and the University of
Neurocognitive Disturbance in OSA 72
Western Australia. All participants provided informed consent to participate in the
study.
Participants
Community Sample. Between January 2011 and September 2011, 356
participants were recruited from the wider community through community and
University notices, community talks, radio announcements, fundraising events, the
University of Western Australian and the West Australian Participant Pool (a panel
of community volunteers aged 50+). Participants were 44.4±22.3 years (range 17.4
to 100.9 years), BMI was 25.3±6.3 (range 15.5 to 62.1) and 248 (70%) were female.
Of these participants 28 reported a diagnosed sleep condition, 11 of which reported
OSA (only 3 reported this to be treated), 3 reported Restless Legs Syndrome and 1
reported Sleep Paralysis.
Clinical Sample. Between March 2009 and July 2011, 679 participants were
recruited through the West Australian Sleep Disorders Institute as they came into the
sleep clinic for overnight assessment of a sleep disorder. Participants were 50.3±22.3
years (range 17.0 to 82.5 years), BMI was 33.03±7.42 kg.m2 (range 16.2 to 61.4),
AHI was 34.8±29.7 events/hr (range 0 to 193) and 305 (44%) were female. Of these
participants 665 later received a diagnosis of OSA, none of whom were treated at the
time of completing this questionnaire.
Questionnaire
The ESS (Johns, 1991) is an eight-item, self-report scale in which
respondents rate their perceived likelihood of dozing in a variety of everyday
Neurocognitive Disturbance in OSA 73
situations. The scale response options range from 0 – “would never doze” to 3 –
“high chance of dozing”, with an ESS score >10 representing excessive sleepiness
(Johns & Hocking, 1997). The measure has strong internal consistency (Cronbach’s
alpha 0.74 – 0.88) (Johns, 1992, 1994; Smith et al., 2008), indicating that the items
of the ESS measure a similar construct. The ESS also exhibits good test-retest
reliability ≥0.8218, indicating that without treatment an individual score will remain
stable when retested.
Statistical analyses
Data screening and descriptive analysis were performed using SPSS 19.0 for
Windows (IBM, 2012b). The factor structure of the ESS was evaluated using AMOS
18.0 for Windows Version 2.0 (IBM, 2012a).
Data screening. Data were screened for missing values. Little’s MCAR test
demonstrated that missing data were missing completely at random in both the
community, X2(21) = 27.02, p = .17, and the clinical sample, X2(80) = 67.70, p =
.84. The proportion of missing values was less than 5% in both samples, thus,
Expectation Maximisation in SPSS was used to replace missing values, before
conducting factor analysis in AMOS.
Confirmatory Factor Analysis. Confirmatory Factor Analysis was used to
compare the 1-factor (Figure 2) and 3-factor (Figure 3) ESS structures. The models
were estimated using Maximum Likelihood (ML) estimation. The fit of the models
was compared through examination of Chi square difference tests and incremental
model fit indices (Normed Fit index, NFI; Incremental Fit Index, IFI; Comparative
Fit Index, CFI; Goodness-of-fit index, GFI; Adjusted Goodness of Fit Index, AGFI;
Neurocognitive Disturbance in OSA 74
Standardized Root Mean Squared Residual, SRMR; and Root Mean Square Error of
Approximation, RMSEA). Particular emphasis was placed on X2 adjusted for
degrees of freedom (X2 / df), the RMSEA and associated confidence intervals, the
SRMR and the CFI, as the power and robustness of these particular indices has
previously been demonstrated (Hu & Bentler, 1999; Iaccobucci, 2010; MacCallum &
Austin, 2010). As fit indices cannot be compared across models (Iaccobucci, 2010),
model cross-validation indices (Akaike’s Inclusion Criterion, AIC; Consistent
Akaike’s Inclusion Criterion, CAIC; and Expected Cross Validation Index, ECVI)
were calculated in order to compare models, as determined by published cut-offs
(See Table 7) (Hu & Bentler, 1999).
Neurocognitive Disturbance in OSA 75
Figure 2.
Proposed 1-factor structure of the ESS
Sleepiness
Item 1
Sitting and reading
Item 2
Watching television
Item 3
Sitting inactive in a public
Item 4
As a passenger in a car
Item 5
Lying down to rest
Item 6
Sitting and talking
Item 8
In a car, while stopped
Item 7
Sitting quietly after lunch
Neurocognitive Disturbance in OSA 76
Moderate
somnificity
Item 1
Sitting and reading
Item 2
Watching television
Item 3
Sitting inactive
Item 4
As a passenger in a car
for an hour without a
Item 5
Lying down to rest
ithe afternoon
Item 6
Sitting and talking
Item 8
In a car, while stopped
traffic
High
somnificity
Low somnificity
Item 7
Sitting quietly after lunch
(you’ve had no alcohol)
Figure 3.
Proposed 3-factor structure of the ESS
Neurocognitive Disturbance in OSA 77
Results
Mean scale scores for ESS total score and each of the 3 factors (e.g. low
somnificity score = average of Item 6 + Item 8 scores) for each sample are shown in
Table 6. As expected, higher scores were obtained from the clinical sample for total
ESS, low and moderate somnificity scores (p < .001) but not for high somnificity
scores, which were similar for both samples. These results indicate that the items
most endorsed were the high and moderately sleepy items, and that of these items the
clinical group endorsed these more than the community group. However, there was
no significant difference between how often the low somnificity items were endorsed
in the clinical and community groups. This indicates that these items are endorsed as
often in the community group as they are in the clinical group. The relationships
between the scores for each ESS item from the clinical and community samples are
presented as a correlation matrix in Appendix 2.
Table 6.
ESS total score, subscale scores for each of the 3 somnificity constructs and t-
test results between the community and clinical samples.
Scale total
Community
sample M±SD
N = 356
Clinical
sample M±SD
N = 693
Sig.
ESS Total score (max. 23) 7.2±4.2 10.0±53 0.001
Low somnificity score (max. 3) 0.1±0.3 0.3±0.6 0.001
Medium somnificity score (max. 3) 0.8±0.7 1.2±0.9 0.001
High somnificity score (max. 3) 1.5±0.7 1.9±0.8 0.724
Neurocognitive Disturbance in OSA 78
Comparisons of the 1 and 3 factor solutions were conducted using adjusted
X2, incremental fit indices and model comparison statistics. The Chi square statistic
for the community sample was significant for the 1-factor, X2(20) = 100.06, p <
.001, and 3-factor, X2(17) = 50.23, p < .001 models. Likewise, the Chi square
statistic for the clinical sample was significant for the 1-factor, X2(20) = 177.30, p <
.001, and 3-factor, X2(17) = 67.61, p < .001, models. However, chi square is highly
affected by sample size and, as such, becomes unstable with moderate to large
samples (Byrne, 2010; Hu & Bentler, 1999; Iaccobucci, 2010; MacCallum & Austin,
2010). Due to the sensitivity of X2 to sample size it has been argued that a model
demonstrates reasonable fit if X2 is ≤ 3 when adjusted by degrees of freedom (X2
/df) (Hu & Bentler, 1999; Iaccobucci, 2010). For both the community and clinical
samples the 3-factor model adjusted X2 was close to or less than 3 (2.95 and 3.98
respectively) whilst the 1-factor model exceeded 3 in both the community and
clinical samples (5.01 and 8.87 respectively). Accordingly, and as recommended, we
report incremental fit statistics (Table 7) (Byrne, 2010; Hu & Bentler, 1999;
Iaccobucci, 2010).
Neurocognitive Disturbance in OSA 79
Table 7.
Fit indices and model statistics for the 1- and 3 factor models in the community and
clinical samples
Community sample
(n=356)
Clinical sample
(n=679)
Fit Index 1-factor 3-factor 1-factor 3-factor
Normed fit index (NFI)1 0.85 0.92 0.92 0.97
Incremental fit index (IFI)1 0.88 0.95 0.92 0.98
Comparative fit index (CFI)1 0.87 0.95 0.92 0.98
Goodness-of-fit index (GFI)1 0.93 0.97 0.94 0.98
Adjusted goodness-of-fit index
(AGFI)1
0.88 0.93 0.89 0.95
Standardized root mean squared
residual (SRMR)2
0.06 0.04 0.05 0.03
Root mean square error of
approximation (RMSEA)3 (lower and
upper confidence intervals)
0.11
(0.09 to
0.13)
0.07
(0.05 to
0.10)
0.11
(0.09 to
0.12)
0.07
(0.05 to
0.08)
Akaike’s Inclusion Criterion (AIC)4 132.06 88.20 209.30 105.61
Consistent Akaike’s Inclusion
Criterion (CAIC)4
210.06 180.86 297.95 210.89
Expected Cross Validation Index
(ECVI)4 (lower and upper confidence
intervals)
0.37
(0.29 to
0.47)
0.25
(0.20 to
0.32)
0.30
(0.25 to
0.37)
0.15
(0.12 to
0.19)
Note. 1 Ideal value ≥ .95; 2 Ideal value ≤ .08; 3 Ideal value ≤ .06; 4 lowest value is the best model (Hu & Bentler, 1999).
Neurocognitive Disturbance in OSA 80
Results presented in Table 7 demonstrate that the 3-factor model accounts for
at least as much variance as the 1-factor model. Firstly, the magnitude of all fit
indices presented in Table 7 was greater for the 1-factor model than the 3-factor
model in both the clinical and population samples. Secondly, for the majority of fit
indices the 3-factor model met or exceeded recommended cut off values in both
samples, whereas the 1-factor model did not. Finally, the validation indices (AIC,
CIAC, ECVI) indicated that the 3-factor model had the lowest values in all instances
indicating this to be the best fit to the data.
Discussion
This study shows that the ESS can provide measurements of situational
somnificity at three different levels, in addition to its commonly used single measure
of sleepiness. The term somnificity reflects the facets of sleepiness encompassing the
broad features of a person’s posture, activity and environment that facilitate sleep-
onset for a majority of people, the majority of the time (Johns, 2002). In the context
of the ESS, a highly somniforic situation is one that induces sleep easily (e.g. lying
down in the afternoon), whilst a low somniforic situation does not normally do so
(e.g. waiting at traffic lights) (Johns, 2002).
An example of the clinical relevance of such a finding is seen in those
individuals who do not achieve a clinically sleepy total ESS score but concede to
falling asleep under conditions that do not normally induce sleepiness. In the clinical
sample, of those who obtained an ESS total score of <10, 5.9% conceded that there
was a slight chance they would doze “whilst sitting and talking to someone” (Item 6)
and 4.4% said there was a slight chance of dozing “whilst in a car stopped in traffic”
Neurocognitive Disturbance in OSA 81
(Item 8). Based on the total ESS score alone, these participants would not normally
be highlighted as having a problem with sleepiness. However, should the construct
scores be available, a high score on the low somnificity scale could raise concern
about the individual’s sleepiness, as they are falling asleep under unusual and
potentially hazardous circumstances.
In addition to its clinical applications, the concept of somnificity is important
for developing an accurate model of sleep processes. Understanding whether or not
the feeling of sleepiness, or even sleep, can be achieved by a particular person in a
particular situation depends upon the characteristics of both the person and the
environment. Understanding the influence of the environment upon our wake and
sleep drives is critical in the diagnosis and treatment of sleep disorders. For example,
cognitive-behavioural therapy for insomnia acknowledges the potentially important
role of environmental sleep cues and incorporates environmental changes into sleep
hygiene practices (Perlis, Jungquist, Smith, & Posner, 2005).
The capacity of the ESS to provide potentially useful information on multiple
levels of somnificity is due to its design, as it contains a number of items that exist
on a continuum of most somniforic to least somniforic (Johns, 1994). Using
normalized factor loadings for four different samples, Johns (Johns, 1994)
demonstrated that items 1, 2 & 5 were highly sleep inducing, 3, 4 & 7 were
moderately sleep inducing and 6 & 8 the least somniforic. The findings of the
present study provide convergent evidence that the ESS can be conceptualised as
measuring three levels of somnificity.
Neurocognitive Disturbance in OSA 82
Previous research has examined the factor structure of the ESS. Many have
found, as did we in both samples, that the fit of the one-factor model was not ideal
(Heaton & Anderson, 2007; Smith et al., 2008). Whilst psychometric evaluation of a
scale aids development and refinement of the measurement of theorized constructs
(Floyd & Widaman, 1995), previous evaluations of the ESS have taken, in our view,
problematic approaches to dealing with model miss-fit. Either they have proposed
the abolition of items from an already very small scale (Smith et al., 2008), utilized
niche populations (Heaton & Anderson, 2007), or utilised less than optimal statistical
techniques (Heaton & Anderson, 2007).
For example, Smith et al (Smith et al., 2008) examined the validity of a
single-factor ESS solution and found that a re-specified model, omitting Items 6 and
8 provided a superior fit to the data. Although deletion of problematic items is a
commonly accepted technique for attaining construct unity (Floyd & Widaman,
1995), this can lead to a loss of information. In particular, the loss of Items 6 (sitting
and talking to someone) & 8 (whilst stopped at traffic lights) would lead to a loss of
information about situations that a majority of people do not find sleep inducing.
Identifying those individuals who do find these situations sleep inducing may be
crucial, however, for recognizing those people for whom sleepiness is particularly
hazardous or problematic. Results from the present paper demonstrate that all items
contribute to an individual’s sleepiness score and, furthermore, provide information
on the type of situations an individual finds most sleep inducing.
A similar analysis conducted by Heaton and Anderson (Heaton & Anderson,
2007) examined workplace risks for long-haul truck drivers. Instead of abolishing
Items 6 & 8, these authors demonstrated that a 2-factor structure with items 6 & 8
Neurocognitive Disturbance in OSA 83
falling onto a second factor offered a superior fit to the 1-factor solution. To extract
factors, Heaton and Anderson (2007) utilized principle components analysis (PCA)
and eigenvalues greater than 1. Heaton and Anderson (2007) defined factor 1 as
“representative of activities that may be solitary and encompass little interaction or
variation in interaction with environment or other people” (pp. 185) and factor 2 as
“activities that may involve more cognitive stimulation and interaction with
environment or other people” (pp.185). However, these factors did not cleanly
separate into 2 distinct factors and there were problematic cross loadings for Items 3
(sitting in a public place) and 7 (sitting quietly after lunch). Heaton and Anderson
(2007) grouped these items with Factor 1, basing their decision on logic rather than
factor loadings. The present findings may explain some of the cross-loadings in the
2-factor design of the ESS proposed by Heaton and Anderson (2007).
With respect to statistical techniques, many previous factor analytic studies
of the ESS cite the Kaiser criterion of eigenvalues greater than 1.0 as the primary
reason to extract a single factor ESS solution (Chen et al., 2002; Johns, 1994).
Unfortunately, this method frequently results in over and under factoring; resulting
in the extraction of too many or too few constructs (Green, Lissitz, & Mulaik, 1977).
Previous studies have also cited the use of Cronbach’s alpha in support of the
homogeneity of the scale (Chen et al., 2002; Johns, 1992). However, high internal
consistency does not necessarily equal high scale homogeneity.28
Of the factor analytic studies published, all have demonstrated some low
factor loadings and, to date, no studies have attempted to contrast the traditional 1-
factor model of sleepiness with a 3-factor model of somnificity using confirmatory
factor analysis. The present research expands upon our previous understanding of the
Neurocognitive Disturbance in OSA 84
ESS by utilizing optimal statistical techniques, utilising the ESS scale in its entirety,
and studying the scale in two large samples.
There are two common pitfalls of using CFA, namely confirmation bias and
the generalizability of results (MacCallum & Austin, 2010). Confirmation bias is an
unfair liking for the model the researcher believes accounts for the most variance in
the data, leading a researcher to evaluate only the fit of the preferred model
(MacCallum & Austin, 2010). Utilizing a comparative model strategy by evaluating
the fit of the traditional model (1-factor) to the hypothesised model (3-factors), as
used in the present paper, lends some protection against confirmation bias
(Iaccobucci, 2010; MacCallum & Austin, 2010). Additionally, often the results
produced in CFA are only generalizable to the population tested (Iaccobucci, 2010).
In order to address this issue the present research assessed the fit of the proposed
models in two distinct samples, making the results more widely generalizable
(Iaccobucci, 2010).
Conclusions
The results of this study support a multidimensional conceptualisation of
sleepiness. Specifically, they indicate that the ESS can be used both as a measure of
overall sleep propensity and as a combination of more specific measures assessing an
individual’s sleep propensity under given situations. This second conceptualisation
of the ESS scale may be useful for more precisely identifying sources of excessive
sleepiness in non-clinical samples. The immediate concerns of an individual
experiencing disordered sleep are likely to relate to the day-to-day consequences of
their condition. As such, it is extremely important to understand the various factors
that contribute to daytime sleepiness, both in terms of ensuring the accurate
Neurocognitive Disturbance in OSA 85
assessment and diagnosis of sleep conditions and addressing the patient’s primary
concerns.
Neurocognitive Disturbance in OSA 86
CHAPTER 5: RELATIONSHIP BETWEEN OSA SEVERITY AND
COGNITION
Preface
Studies 1 and 2 (presented in Chapters 2 and 4) examined the pattern of
deficits in executive function, and the factor structure of the Epworth Sleepiness
Scale (ESS), respectively. Chapter 2 revealed that individuals with OSA experience
widespread executive dysfunction that improves with treatment. Furthermore,
published literature indicates that attention and memory are likewise impaired in
individuals with OSA. The findings from Chapter 4 revealed that sleepiness, as
measured by the ESS, was best captured as a 3-factor model measuring low, medium
and high situational somnificity.
This chapter uses this information to examine the relationship between the
domains of attention, memory and executive function and nocturnal disturbance
(hypoxia and sleep fragmentation), whilst controlling for the inter-individual factors
of sleepiness, age and premorbid IQ.
This chapter was submitted to Sleep and Breathing, presently, the authors are
replying to reviewer comments: Olaithe M, Skinner T., Hillman D., Eastwood P.,
Bucks R. The relationship between cognition and sleep apnea in OSA: a call to
arms. Sleep and Breathing (Submitted and under review, December, 2013).
Neurocognitive Disturbance in OSA 87
Abstract
Introduction: Obstructive Sleep Apnea (OSA) is a common disorder that is associated
with impaired attention, memory and executive function. However, the mechanisms
underlying such dysfunction are unclear. To determine the influence of sleep
fragmentation and hypoxia, this study examined the effect of sleep fragmentation and
hypoxia on cognition in OSA, while controlling for potentially confounding variables
including sleepiness, age and premorbid intelligence.
Method: Participants with and without OSA (N = 150) were recruited from the
general community and a tertiary hospital sleep clinic. All underwent comprehensive,
laboratory-based polysomnography (PSG) and completed assessments of cognition
including attention, short-term & long-term memory and executive function.
Structural Equation Modelling (SEM) was used to construct a theoretically-driven
model to examine the relationships between hypoxia and sleep fragmentation, and
cognitive function.
Results: Although after controlling for IQ, increased sleep disturbance was a
significant predictor of decreased attention (p=0.04) and decreased executive function
(p=0.05), controlling for age removes these significant relationships. No significant
predictors of memory function were found.
Conclusions: The mechanisms underlying the effects of OSA on cognition remain to
be defined. Implications are discussed in light of these findings.
Neurocognitive Disturbance in OSA 88
Cognition and nocturnal disturbance in OSA; the importance of
accounting for age and premorbid intelligence
Obstructive Sleep Apnea (OSA) is a common sleep disorder (T. Young et al.,
2002) characterized by repeated upper airway collapse resulting in intermittent
hypoxia and arousals from sleep (Kryger, 2010). Previous studies, comparing
individuals with and without OSA, have shown that OSA is associated with impaired
cognitive function, particularly in the domains of attention, memory, and executive
function. In the attention domain, vigilance appears to be most affected (Beebe et al.,
2003; Bucks et al., 2012; Fulda & Schulz, 2003), while in the memory domain,
impairment is in delayed visual and verbal memory (Aloia et al., 2004; Bucks et al.,
2012; Wallace & Bucks, 2012), and in visuospatial/constructional memory (Bucks et
al., 2012). Executive function impairments have been demonstrated in Shifting,
Updating, Inhibition, Generativity, and Fluid reasoning (Beebe et al., 2003; Bucks et
al., 2012; Fulda & Schulz, 2003; Olaithe & Bucks, 2013; Saunamäki & Jehkonen,
2007). Several domains of cognition appear unaffected by OSA, including language
abilities (Beebe et al., 2003; Bucks et al., 2012), immediate visual and verbal
memory (Wallace & Bucks, 2012), visuospatial learning (Wallace & Bucks, 2012),
and psychomotor functions (Bucks et al., 2012).
The mechanisms underlying cognitive dysfunction in OSA remain undefined.
In 2002, Beebe and Gozal (Beebe & Gozal, 2002) proposed a conceptual framework
based around critical roles for sleep fragmentation and nocturnal hypoxia in the
development of cognitive dysfunction in individuals with OSA. In their model, sleep
is viewed as a necessary restorative process, regulating processes including
reinforcing foundations for learning and memory (Poe et al., 2010) and modulating
Neurocognitive Disturbance in OSA 89
neuroendocrine demands (Everson, 1995). Disruption of these processes due to sleep
fragmentation leads to an inability of the body to return to a balanced state,
impairing neural function. The model also postulates that blood gas abnormalities
(i.e., hypoxia) due to periodic obstructed breathing compounds this damage. A
proposal of this model is that hypoxia and sleep fragmentation contribute equally to
impaired cognition in OSA, however it remains unclear whether or not this is
correct.
Substantial evidence exists for a negative impact on cognition by sleep
fragmentation and/or sleep deprivation (Alhola & Polo-Kantola, 2007; Durmer &
Dinges, 2005b; Stepanski, 2002; Verstraeten et al., 2004). Indeed, some authors
argue that sleep fragmentation is the key mechanism underlying impaired cognition
in OSA (Verstraeten & Cluydts, 2004; Verstraeten et al., 2004). Of note, is the
finding that the nature of cognitive deficits in sleep deprived individuals is similar to
those individuals with OSA (Alhola & Polo-Kantola, 2007; Durmer & Dinges,
2005b; Verstraeten et al., 2004) Particular impairment in attention and executive
function are prominent amongst them (Alhola & Polo-Kantola, 2007; Durmer &
Dinges, 2005b; Verstraeten et al., 2004). It has also been noted that deficits in
attention in individuals with OSA are associated with indices of sleep fragmentation
(Ayalon, Ancoli-lsrael, Aka, McKenna, & Drummond, 2009).
Other research suggests that intermittent hypoxia experienced throughout the
night by individuals with OSA is primarily responsible for irreparable neural damage
and lasting cognitive impairments (Beebe & Gozal, 2002; Beebe et al., 2003). Such
permanent damage might explain the finding that, even with effective treatment (viz.
abolition of hypoxia and sleep fragmentation), individuals with OSA retain some
Neurocognitive Disturbance in OSA 90
decrements in memory and executive functioning (Bucks et al., 2012; Olaithe &
Bucks, 2013; Wallace & Bucks, 2012). Thus, while there is considerable evidence
that both sleep fragmentation (Bartlett et al., 2004; Chiang, 2006; Gale & Hopkins,
2004; Hopkins, Kesner, & Goldstein, 1995; Naismith, Winter, Gotsopoulos, Hickie,
& Cistulli, 2011) and hypoxia (Aloia et al., 2004; Bucks et al., 2012; Saunamäki et
al., 2010) lead to cognitive dysfunction in OSA, their relative importance remains
unclear.
In addition to hypoxia and sleep fragmentation, other inter-individual factors
including age (Naismith et al., 2004), premorbid intelligence and sleepiness should
be considered in any assessment of cognitive function in individuals with OSA. This
is because those individuals with OSA who have high intelligence may exhibit intact
performance on neuropsychological assessments, as they can recruit additional
cognitive resources or utilise cognitive strategies that aid in performance, acting as a
‘buffer’ for brain insult and injury (i.e., cognitive reserve) (Alchanatis et al., 2005).
Similarly, individual differences in age and subjective sleepiness in OSA could
modify performance on neurocognitive tests because age independently affects
cognition (Alchanatis, Zias, Deligiorgis, Chroneou, et al., 2008), and excessive
sleepiness lessens the ability to direct cognitive resources to attend to the task at
hand (Durmer & Dinges, 2005a, 2005b; Guilleminault & Brooks, 2001; Naismith et
al., 2004).
Despite the wealth of published evidence, most studies of the cognitive burden
of OSA have compared OSA to non-OSA participants (Bucks et al., 2012). By
contrast, few studies have explored the relationship between the severity of OSA
(e.g. as quantified by the Apnoea Hypopnea Index, AHI) and the degree of deficits
Neurocognitive Disturbance in OSA 91
experienced, and none has explored the relative contribution of the two proposed
mechanisms (that is sleep fragmentation AND hypoxia) within the same sample,
whilst controlling for premorbid intelligence, age and daytime sleepiness (Naismith
et al., 2004). Thus, this study is the first comprehensively to test the relationship
between severity of OSA and cognitive deficits. To do this we utilized Structural
Equation Modelling (SEM), a powerful analytic technique that allows testing and
estimation of causal relationships between multiple potential mechanisms of
cognitive dysfunction, as well as exploration of the relative importance of inter-
individual factors of age, premorbid IQ, and sleepiness.
Methods
Participants and Protocol
In total, 150 adults (18 years+) with and without OSA participated: 134 were
patients diagnosed with OSA attending the Sleep Clinic of the West Australian Sleep
Disorders Research Institute, Queen Elizabeth II Medical Centre (WASDRI-QEII),
Western Australia, for diagnostic polysomnography between March 2009 and July
2011. Cognitive assessments were performed with these individuals after diagnosis
but before starting treatment for OSA. To capture the full range of severity of OSA
(from no disease to severe disease) an additional 16 community volunteers were
recruited. These individuals were members of the West Australian Participant Pool
of the University of Western Australia (WAPP-UWA), who completed a set of
questionnaires on sleep health between October 2011 and October 2012, and who
undertook cognitive assessment prior to a sleep study at the Centre for Sleep
Science, at the University of Western Australia. The study was approved by the
Neurocognitive Disturbance in OSA 92
Human Research Ethics Committees of Sir Charles Gairdner Hospital, and the
University of Western Australia. All participants provided written, informed consent.
Measurements
Polysomnography (PSG) required participants to attend a full, overnight
laboratory-based sleep study. Electrodes were attached according to the American
Academy of Sleep Medicine (AASM) recommendations: electroencephalogram
(EEG) electrode, electrooculogram (EOG) channels, electromyogram (EMG) (chin
and legs), oxygen saturation via pulse oximetry (SpO2), electrocardiogram (ECG),
thoracic and abdominal respiratory effort via belts, oral/nasal thermistor, nasal
prongs and position sensor. Studies were analysed and apnea hypopnea index (AHI)
and arousal index (ArI) determined according to standard guidelines (AASM, 2007).
Attention and Memory were assessed using the Cognitive Drug Research
System (CDR; United BioSource Corporation), a 30-minute computerized battery of
cognitive assessments (Van Den Goor et al., 2008; Wesnes, 2000). All CDR tasks
were administered on a 15-inch screen laptop computer, and included assessment of
Attention (digit vigilance, simple and choice reaction time), Long Term Memory
(delayed picture recognition, delayed word learning, delayed word recognition) and
Short Term Memory (digit recall, visuo-spatial recall). Responses were provided
either verbally or via a YES/NO response box.
Executive Function was assessed using the Clock Drawing Task (CLOX 1 &
2) (Royall et al., 1998), the Controlled Oral Word Association task (COWA), and the
Trail Making Test (TMT; Trails A and B) (AASM, 1999; Partington & Leiter, 1949;
Reitan, 1958).
Neurocognitive Disturbance in OSA 93
Pre-morbid Intelligence was assessed using the National Adult Reading Test -
2 (NART-2), is a widely accepted method of estimating premorbid IQ from current
capacity, by testing reading capacity of irregular nouns (Nelson & Willison, 1991).
The number of errors is used to estimate IQ using a regression equation, with fewer
errors being associated with higher premorbid IQ. Irregular word reading ability has
been shown both to correlate highly with full scale IQ (Nelson & Willison, 1991)
and to be resistant to decline due to dementia (McGurn et al., 2004).
Subjective Sleepiness was assessed using the Epworth Sleepiness Scale (ESS)
(Johns, 1991) including its separation into three factors, of low, moderate, and high
situational somnificity (Olaithe, Skinner, Clarke, Eastwood, & Bucks, 2012).
All cognitive assessments were administered by graduate psychologists at the
University of Western Australia, and all sleep study measurements were obtained
and scored by trained PSG Technicians.
Analyses
Data screening and descriptive analysis were performed using SPSS 20.0 for
Windows (IBM, 2012b).
Characterising hypoxia and sleep fragmentation. The degree of hypoxia was
operationalised in two ways: (i) mean overnight SpO2; and (ii) minimum overnight
SpO21. The degree of sleep fragmentation was also represented in two ways: (i) sleep
1 Other indices of sleep fragmentation (e.g., TST, ArI during REM) and the following measures of hypoxia: CT90, Lowest SaO2 during REM, Mean SaO2 during REM) were considered and tested, none accounted for sufficient variance in cognitive performance to be included.
Neurocognitive Disturbance in OSA 94
efficiency, the ratio of time spent asleep (total sleep time) to the amount of time
spent in bed; and (ii) proportion of arousals not associated with a 3% oxygen
desaturation (Total ODI3) (Sulit, Siorfer-Isser, Kirchner, & Redline, 2006), the latter
being the number of times an individual’s SpO2 decreased to below 3% of baseline2.
The purpose of this index was to model arousals not associated with a significant
drop in oxygen, as significant respiratory events generally occur with an arousal
making these two factors highly correlated.
Data screening. Little’s MCAR revealed that data were missing completely at
random, X2(485) = 10.19, p = 1.00. The proportion of missing values was less than
10%, thus Expectation Maximisation was used to replace missing values (Cohen &
Cohen, 1983). Data were assessed for normality. Many variables violated normality
assumptions, hence Generalised Least Squares (GLS) was used to calculate estimates
of model fit (Lei & Lomax, 2005). Furthermore, incremental fit indices are reported,
as these have been shown to be less sensitive to non-normality (Lei & Lomax, 2005).
Confirmatory Factor Analysis. CFA is a confirmatory form of SEM that can be
used to confirm a theorised model. SEM is an extension of multiple regression
designed to test a set of hypothesised relationships between variables, that are
estimated simultaneously (Ullman & Bentler, 2012). It provides a mechanism
through which to examine relationships between hypothesized constructs, whilst
controlling for individual differences, such as premorbid intelligence and sleepiness,
and accounting for measurement error. SEM is particularly useful when examining
Neurocognitive Disturbance in OSA 95
hypothesized constructs such as memory, attention, and executive function, as it can
account for measurement error that naturally exists between the ‘pure’ construct and
its measurement providing a stringent test of the latent structure (Byrne, 2010;
Iaccobucci, 2010; MacCallum & Austin, 2010). The weights presented in the figures
represent standardized regression weights, and Appendix 3 presents a correlation
matrix of all variables included in the final model.
CFA was used to evaluate the theoretical models (Figure 4 and 5). The fit of
the models was compared through examination of the Chi square difference test, and
incremental model fit indices. ‘Good fit’ indicates that the hypothesised model
generates a sample covariance matrix that is similar to the observed covariance
matrix inherent in the data set; that is the model is a good explanation of the variance
in the observed data (Schreiber, Stage, King, Nora, & Barlow, 2006). Good fit is
indicated by a non-significant Chi square value, indicating that the observed and
hypothesised models are similar, and fit indices are within recommended guidelines
(Table 8 foot note). Particular emphasis is placed on the values of the Comparative
Fit Index, CFI; Standardized Root Mean Squared Residual, SRMR; and Root Mean
Square Error of Approximation, RMSEA, as the power and robustness of these
particular indices has previously been demonstrated (Byrne, 2010; Hu & Bentler,
1999; Iaccobucci, 2010; MacCallum & Austin, 2010). The factor structure of the
models was evaluated using AMOS 18.0 for Windows Version 2.0 (IBM, 2012a).
The correlation matrix is provided in Appendix 3.
Four models were analysed. The first model examined the interrelationships
between hypoxia/sleep fragmentation and cognitive function, controlling for
premorbid IQ. The second model repeated this analysis, but controlled for premorbid
Neurocognitive Disturbance in OSA 96
IQ and age. A third model examined the interrelationships between AHI and
cognitive function by replacing the measures of hypoxia and sleep fragmentation
with AHI. This was performed because AHI is the primary measure of OSA disease
severity used in current clinical practice (Kribbs et al., 1993; Tsai, 2010) and the
utility of this single measure to predict cognitive dysfunction is disputed (Kribbs et
al., 1993; Tsai, 2010). The fourth model assessed the role of attention in mediating
cognitive deficits in the domains of short and long-term memory and executive
function. This model was examined because cognitive deficits OSA have been
proposed to be due to attention difficulties as a consequence of sleep fragmentation
(Verstraeten et al., 2004).
Results
Descriptive data and preliminary analyses
Participants were aged 52.4±13.1 years (range 19 to 82 years), BMI was
33.0±7.8 kg/m2 (range 15.5 to 57.5), AHI was 37.1±27.5 events/hr (range 0 to 154),
had received 11.9±3.1 years (range 6 to 23) of education, and 69 (46%) were female.
Descriptive statistics for cognitive and OSA indices are provided in Table 8.
Neurocognitive Disturbance in OSA 97
Table 8.
Descriptive statistics for sleep and cognitive measures for the whole sample.
Mean Standard
Deviation
Min. Max.
ArI, /hr 38.3 21.5 0.0 114.5
Total ODI3 132.6 135.4 0.0 718.0
Arousals proportion ODI3 4.2 4.6 0.0 28.6
Sleep Efficiency, % 75.4 12.1 38.6 95.8
Mean SaO2, % 92.8 3.3 73.0 98.0
Lowest SaO2, % 83.6 8.7 29.0 96.0
ESS Total Score 9.5 4.9 0.0 21.4
Somnificity Factor Low (ESS) 0.0 1.0 -0.6 3.6
Somnificity Factor Moderate (ESS) 0.0 1.0 -1.3 2.4
Somnificity Factor High (ESS) 0.0 1.0 -2.5 1.5
CLOX Ratio 1.7 2.1 -4.0 10.0
Trails Ratio 2.5 0.8 1.1 6.0
Verbal Fluency 37.0 12.5 3.0 68.0
Immediate Spatial Recall 93.5 11.8 12.5 100.0
CDR Word Learning 0.9 1.9 -7.0 7.0
Delayed Word Recognition 0.6 0.2 -0.7 0.9
Delayed Picture Recognition 0.7 0.2 0.2 1.1
Choice Reaction Time 96.9 2.6 84.0 100.0
Digit Vigilance Score 96.5 5.1 75.6 100.0
Note. Estimated Premorbid IQ, based on the NART-2, years education, age and occupation (population Mean
100±15); Arousals proportion ODI3 = The number of arousals as a proportion of the number of times an individual desaturated
3% below baseline; Sleep Efficiency = is the ratio of time spent asleep (total sleep time) to the amount of time spent in bed.
Neurocognitive Disturbance in OSA 98
Compared to the clinical sample (n = 134), the community sample (n = 16) was
younger (p = 0.001), and had more years of education (p = 0.003). As anticipated
they had lower BMI (p = 0.003) and lower AHI (p = 0.001). There were no
differences between the groups in estimated premorbid intelligence (p = 0.23). For
descriptive statistics on these variables, and for the sample as a whole see Table 9.
Age and education were accounted for within the SEM analysis.
Neurocognitive Disturbance in OSA 99
Table 9.
Descriptive statistics provided for the clinical and community samples, and for the sample as a whole.
Community sample Clinical sample Whole sample
Mean Standard
Deviation
Min. Max. Mean Standard
Deviation
Min. Max. Mean Standard
Deviation
Min. Max.
Age 40.3 14.7 19 67 53.9 12.1 25 82 52.4 13.1 19 82
BMI 24.4 4.6 16.5 34.3 34.1 7.5 15.5 57.5 33.0 7.8 15.5 57.5
Gender 44% male 7 male 55% male 74 males 54% male 81 males
Education 12.7 2.9 10.0 20.0 11.6 3.0 6.0 23.0 11.9 3.1 6.0 23.0
Estimated Premorbid IQ 112.9 5.5 100.0 121.6 110.4 8.0 86.3 131.1 110.7 7.8 86.3 131.1
AHI 2.2 3.9 0.0 12.8 41.3 26.1 8.1 154.8 37.1 27.5 0.0 154.8
Note. AHI = Apnoea Hyponoea Index; BMI = Body Mass Index; Estimated Premorbid IQ based on Crawfords equation using the NART.
Neurocognitive Disturbance in OSA 100
Model I: Hypoxia/sleep fragmentation vs cognitive function
Preliminary assessment of model weights showed that neither sleepiness, nor
somnificity factors (Olaithe et al., 2012) were related to any cognitive domain (p >
.05 for all) (Appendix 3, Table 2), whilst premorbid intelligence was related to the
cognitive domains (p < .05) (Figure 4). Thus, sleepiness/somnificity was removed
from subsequent models whilst premorbid intelligence was retained.
Neurocognitive Disturbance in OSA 101
Attention
Executive
Function
Short Term
Memory
Long Term
Memory
Digit Vigilance Score
Immediate Spatial Recall
Immediate Numeric Recall
CDR Word Learning
Word Recognition
Delayed
Picture Recognition
Delayed
Verbal Fluency
Trails Ratio
Choice Reaction Time
Sleep
Fragment
Hypoxia
Arousals as a proportion of
ODI3
Sleep Efficiency
Mean SpO2
Lowest SpO2
CLOX Ratio
-0.32
0.91
0.67
-0.83
0.88**
0.89
1.66
1.54
0.64
0.79
0.83**
0.74
0.74
0.36
0.68
0.44
-0.47
0.34
0.80
-0.45
-0.16
Estimated IQ
0.68
0.36
0.13
0.41 0.45
-0.02
Figure 4. Model I - The Sleep fragmentation (Sleep Fragment) and Oxygen Desaturation (Hypoxia) Model showing the relationship between
nocturnal disturbance and cognitive constructs; Attention, Long Term Memory, Short Term Memory, and Executive Function, controlling for IQ. All weights are presented as standardised regression weights. Note. This model was also explored controlling for daytime sleepiness (using the ESS). ESS scores produced poor fit and were removed.
Neurocognitive Disturbance in OSA 102
The final model (Figure 4) was a good account of the relationships between sleep
fragmentation (i.e., proportion non-ODI3 arousals and sleep efficiency) and hypoxia
(i.e., mean and lowest SpO2) and cognitive function (Figure 4). Specifically, good
model fit was indicated by a non-significant Chi square statistic, X2(79) = 91.52, p =
0.16, and incremental fit indices were within recommended guidelines (Table 10).
The strength of the interrelationships between sleep fragmentation, hypoxia
and each cognitive factor (short and long-term memory, attention and executive
function) was assessed by examination of the beta weights (Figure 4). Because the
sleep fragmentation and hypoxia factors were highly correlated (r = 0.88 some paths
from these factors to cognition had beta weights > 1.0 (Deegan, 1978; Joreskog,
1999), the model fit was assessed when sleep fragmentation and hypoxia were
combined into a single construct, ‘sleep disturbance’ (Appendix 3, Figure 1). This
model included measures of ArI as a proportion of total ODI3, Sleep efficiency,
Mean and Lowest SpO2 loading onto a latent construct termed ‘sleep disturbance’,
and examined the relationship between these constructs and cognition. Although Chi
square model fit was non-significant, the Akaike Information Criterion (AIC)
indicated that the one factor model (i.e., sleep disturbance) provided a poorer fit to
the data than the original model. Thus, we retained the model separating sleep
fragmentation and hypoxia.
The beta weights of Model I revealed significant relationships between sleep
fragmentation and both attention (r = 1.16, p = 0.04) and executive function (r =
1.08, p = 0.03), after controlling for premorbid intelligence (Figure 4), indicating that
increased sleep fragmentation is significantly and positively related to increased
difficulty with attention and with executive function.
Neurocognitive Disturbance in OSA 103
Model II: Controlling for age
The model above was also examined controlling for age (Figure 5). This
analysis was separately in order to allow examination of the relationships prior to the
removal of age from the model, as age was likely to be strongly associated with
disease duration.
Neurocognitive Disturbance in OSA 104
Attention
Executive
Function
Short Term
Memory
Long Term
Memory
Digit Vigilance Score
Immediate Spatial
Recall
Immediate Numeric
Recall
CDR Word Learning
Word Recognition Delayed
Picture Recognition
Delayed
Verbal Fluency
Trails Ratio
Choice Reaction Time
Sleep
Fragment
Hypoxia
Arousals as a proportion of
ODI3
Sleep Efficiency
Mean SpO2
Lowest SpO2
CLOX Ratio
-0.39
0.93
0.63
-0.70
0.70
0.56
0.87
0.67
-0.11
0.05
0.34
0.21
0.98
0.22
0.76
0.41
-0.37
0.25
0.71
-0.55
-0.07
Estimated IQ
0.76
0.25
0.23
0.41 0.55
-0.07
Age
-0.41
-0.47
-0.40
-0.01
Figure 5.
Model II – The Sleep fragmentation (Sleep Fragment) and Oxygen Desaturation (Hypoxia) Model showing the relationship between nocturnal disturbance and cognitive constructs; Attention, Long Term Memory, Short Term Memory, and Executive Function, controlling for IQ, and age. All weights are presented as standardised regression weights. Note. This model was also explored controlling for daytime sleepiness (using the ESS). ESS scores produced poor fit and were removed.
Neurocognitive Disturbance in OSA 105
This model was a good account of the relationships between sleep fragmentation
(i.e., arousals as a proportion of total ODI3 and sleep efficiency) and hypoxia (i.e.,
mean and lowest SpO2) to cognitive function (Figure 5). Specifically, good model fit
was indicated by a non-significant Chi square statistic, X2(90) = 103.72, p = 0.15 and
incremental fit indices were within recommended guidelines (Table 10).
Neurocognitive Disturbance in OSA 106
Table 10.
Fit indices for both the AHI and sleep fragmentation/hypoxia models are presented.
Fit Index Sleep
fragmentation/hypoxia
with Premorbid IQ
(Model I)
Sleep
fragmentation/hypoxia
with Premorbid IQ,
and age (Model II)
Normed fit index (NFI)1 0.54 0.54
Incremental fit index (IFI)1 0.88 0.90
Comparative fit index (CFI)1 0.84 0.87
Goodness-of-fit index (GFI)1 0.92 0.91
Adjusted goodness-of-fit index
(AGFI)1
0.87 0.87
Standardized root mean squared
residual (SRMR)2
0.08 0.08
Root mean square error of
approximation (RMSEA)3
(lower and upper confidence
intervals)
0.03
(0 to 0.06)
0.03
(0 to 0.06)
Note. 1 Ideal value ≥ 0.95; 2 Ideal value ≤ 0.08; 3 Ideal value ≤ 0.06; 4 Lowest value is indicative of best model (Hu &
Bentler, 1999).
When age was added, however, there was no longer a significant relationship
between either of the sleep disturbance factors and any cognitive factor. That is, the
relationship between sleep fragmentation, attention and executive function was no
longer found.
Neurocognitive Disturbance in OSA 107
Model III: AHI
The model fit was also assessed when hypoxia and sleep fragmentation were
replaced by AHI. For this model (Appendix 3, Figure 2) Chi square was statistically
significant, X2(57) = 86.31, p = 0.007, and incremental fit indices did not meet
required cut points (CFI = 0.39, RMSEA = 0.06, SRMR = 0.08). Such findings
indicate that severity of AHI does not predict level of cognitive performance in OSA.
Model IV: Cognitive dysfunction mediated by attention
A model was also developed to test the hypothesis that cognitive deficits in
executive function and memory are mediated by attention (Appendix 3, Figure 3).
This model would not converge, hence this model provided a poor account of the
data and could not be explored further.
Discussion
This study used structural equation modeling to examine the effects on
cognition of sleep fragmentation and hypoxia in OSA, while controlling for the
potential confounding influences of age, premorbid intelligence, mood and
sleepiness.
Before controlling for age, greater sleep fragmentation significantly predicted
poorer executive function and attention, whereas hypoxia was unrelated to cognition.
After controlling for age, sleep fragmentation joined hypoxia in failing to predict
level of cognitive performance in any domain (attention, language, executive
function, or episodic memory).
Neurocognitive Disturbance in OSA 108
This is not the first study to report no relationship between the severity of
cognitive deficits and the severity of OSA (see studies reviewed by Aloia et al.
(2004), and presented in Table 3 of their publication). Previous studies, using the
AHI, have explained their finding by arguing that AHI is a poor index of OSA
severity, at least as it relates to the cognitive burden of the disorder (Tsai, 2010).
Other authors have suggested that the lack of a relationship is due to the lack of
control of the inter-individual factors of age, premorbid IQ, daytime sleepiness or
attention (Alchanatis et al., 2005; Alchanatis, Zias, Deligiorgis, Liappas, et al., 2008;
Tsai, 2010; Verstraeten et al., 2004).
This study addressed all of these methodological criticisms and still failed to
find a relationship between OSA severity and cognition. What should be made of
such findings?
The simplest explanation for the results of the present and other similar studies
(Aloia et al., 2004; Wallace & Bucks, 2012), is that there is no dose-response
relationship in OSA. That is to say, over a critical threshold there will be neural
damage and cognitive dysfunction, but this will not increase with increasing severity
of sleep fragmentation and hypoxia.
However, given that cognitive performance is related to time spent without
oxygen in other instances of hypoxia (e.g., climbing at altitude (Regard, Oelz,
Brugger, & Landis, 1989), carbon monoxide poisoning (Hopkins & Woon, 2006)),
and longer time exposed to sleep deprivation results in greater cognitive dysfunction
(Durmer & Dinges, 2005a), this account seems unlikely. An alternative hypothesis is
Neurocognitive Disturbance in OSA 109
that hypoxia or sleep fragmentation are not captured appropriately, and/or that the
correct mechanisms of harm are not being captured.
An individual’s level of hypercapnia is closely linked to their level of hypoxia,
and excessive levels of carbon dioxide are found in individuals with OSA (AASM,
2007; Kaw, Hernandez, Walker, Aboussouan, & Mohlesi, 2009). Yet hypercapnia is
not routinely explored in OSA-cognition research, despite being included as a
potential mechanism of harm in the dominant model of cognitive harm in OSA
(Beebe & Gozal, 2002). Further, there is evidence that the degree of hypercapnia
correlates with overall cognitive impairment (Findley et al., 1986).
Alternatively, routine indices of sleep fragmentation (ArI) and hypoxia (SpO2)
may be insufficiently sensitive to the extent of sleep disturbance, or the length and
depth of hypoxia (Asano et al., 2009). An individual with OSA may exhibit many
small fragmentations of sleep architecture, or shallow, lengthy oxygen desaturations,
not captured by standard indices. Novel measures that capture more detail about
sleep fragmentation (Pepin et al., 2005), or describe the length and depth of
desaturation (e.g. Integrated Area of Desaturation) (Asano et al., 2009) may
demonstrate greater sensitivity to a dose-response relationship in OSA.
The impact of inter-individual differences; Age and IQ
The final model showed that, when premorbid IQ was accounted for, more
severe sleep fragmentation, but not hypoxia, was significantly associated with poorer
attention and executive dysfunction, but not with poorer memory. However, adding
age resulted in these relationships becoming non-significant. A simple explanation
for this finding is that age accounts for all of the cognitive deficits seen in OSA. This
Neurocognitive Disturbance in OSA 110
is unlikely, as case-control studies (with participants matched for age) reveal clear
evidence of cognitive deficits in OSA (Bucks et al., 2012). Furthermore, not all
individuals experience age-related change in cognition individuals experience age-
related cognitive decline, nor are all individuals affected equally. Indeed, many older
adults stay cognitively healthy (Hultsch, Hertzog, Small, & Dixon, 1999; Kramar,
Erikson, & Colcombe, 2006; Norbury, Craig, Cutter, Whitehead, & Murphy, 2004).
Given that there is no good measure of disease duration in OSA, and that older
individuals are likely to have experienced OSA for longer, age itself is confounded
with disease duration. By controlling for age, it is possible that the variance of
interest in an exploration of the mechanisms of cognitive harm in OSA has been
discarded. One of the challenges for the field of sleep medicine, therefore, is to
develop sound methods of estimating or measuring disease duration.
The choice of measures of sleep fragmentation and hypoxia in the present
study was guided by indices that are routinely produced by most sleep software, and
that have previously been shown to be related to aspects of cognition (Aloia et al.,
2004). Statistical modelling showed that measures of hypoxia and sleep
fragmentation exhibited a large degree of co-variation, indicating shared underlying
variance, even after calculating an index of arousals independent of 3% oxygen
desaturations. The importance of this covariance was investigated by modifying the
original model such that sleep fragmentation and hypoxia were considered as a
single factor by: (i) loading all measured variables from the latent factors ‘sleep
fragmentation’ and ‘hypoxia’ to a single factor (termed sleep disturbance); and (ii)
replacing sleep fragmentation and hypoxia with AHI. In both cases, the resultant
model fits were not improved relative to the model that utilised separate constructs
Neurocognitive Disturbance in OSA 111
of hypoxia and sleep fragmentation, suggesting that separating out these components
of potential harm was the correct approach.
The present study was also careful to consider the potentially important
confounding influences of premorbid intelligence and sleepiness in assessing
cognitive function in individuals with OSA. The finding that premorbid intelligence
was positively related to cognition was expected, and underscores the importance of
controlling for this when evaluating cognition in OSA (AASM, 1999; Alchanatis et
al., 2005; Ayalon, Ancoli-Israel, Klemfuss, Shalauta, & Drummond, 2006;
Castronovo et al., 2009), In contrast, analysis of subjective sleepiness revealed that
neither ESS total, nor factors of situational somnificity, were related to cognition.
Such findings support studies reporting that subjectively measured sleepiness and
somnificity are not related to cognition in OSA (Incalzi et al., 2004). One reason for
this might be that individuals with chronic sleep restriction, as occurs in OSA, can
become desensitized to the feeling of sleepiness (Durmer & Dinges, 2005b).
The role of attention in mediating cognitive deficits in OSA warranted a
separate analysis, given the proposal that sleep fragmentation might cause impaired
attention which, of itself, could lead to impaired memory and executive function
(Verstraeten, 2007; Verstraeten & Cluydts, 2004; Verstraeten et al., 2004). However,
the resultant model fit was poor, suggesting that attention deficits did not explain the
effect of sleep fragmentation on attention and executive function.
Cognitive damage, but no relationship to disease severity
Despite growing evidence that OSA is associated with cognitive impairment
(Bucks et al., 2012), the findings of the present study, and those of meta-analyses
Neurocognitive Disturbance in OSA 112
and systematic reviews have found no consistent evidence that the degree of sleep
fragmentation or hypoxia is related to the degree of cognitive dysfunction in OSA
(Aloia et al., 2004; Wallace & Bucks, 2012). There are several possible reasons for
this lack of association.
First, it is possible that the cognitive dysfunction in OSA is due to a
mechanism not captured in the models presented in the current study. One such
mechanism could be recruitment of additional brain regions to compensate for poor
cognitive performance. Consistent with this hypothesis, Ayalon et al. (Ayalon et al.,
2006) reported that individuals with OSA recruit additional brain areas, not typically
recruited during a verbal learning task and Castronovo et al. (Castronovo et al.,
2009) reported increased activation in the left frontal cortex, medial precuneus, and
hippocampus in OSA patients whilst performance was at the same level as controls.
Second, individual differences might play a more critical role than previously
credited (Durmer & Dinges, 2005b). Such a concept arises from emerging research
indicating that the relationship between disease severity and cognitive dysfunction is
the product of a multitude of vulnerability and protective factors (Beebe, 2006;
Beebe & Gozal, 2002), of which sleep fragmentation, hypoxia and cognitive reserve
are only three aspects (Casseli, 2008; Durmer & Dinges, 2005b; Zimmerman &
Aloia, 2012). These other facets could include duration of disease (as noted above)
(Casseli, 2008), genetic vulnerability (e.g., apolipoprotein e4 genotype) (Casseli,
2008; Zimmerman & Aloia, 2012), the role of the blood brain barrier (Lim & Pack,
2013), and cerebral blood flow (Kiralti, Demir, Volkan-Salanci, Demir, & Sahin,
2010).
Neurocognitive Disturbance in OSA 113
Third, the present study examined attention, memory and executive function as
whole domains, rather than exploring separate facets of these domains (Olaithe &
Bucks, 2013; Wallace & Bucks, 2012). Cognitive domains, such as executive
function are not unitary constructs, and should be considered umbrella terms for a
range of different cognitive capacities (Fisk & Sharp, 2004; Lezak et al., 2004;
Miyake et al., 2000). For example, executive function can be divided into Shifting,
Updating, Inhibition, Generativity, and Fluid reasoning (Fisk & Sharp, 2004; Lezak
et al., 2004; Miyake et al., 2000). Decomposing each cognitive domain, and then
exploring the contribution of a range of risk and protective factors specific to OSA,
could enable a more targeted analysis of the impact of OSA on cognition.
Last, as mentioned briefly before, measurement of hypoxia through SpO2 may
be a poor proxy measure of the underlying mechanism by which hypoxia causes
harm, namely oxidative stress (Guo, Yan, Qing, Min, & Huan, 2013; Nair, Dayyat,
Zhang, Wang, & Gozal, 2011), particularly given that oximetry measured below
80% SpO2 is not particularly reliable or accurate (Razi & Akbari, 2006). Animal
models suggest that oxidative stress and apoptosis-related neural injury might be
measured more directly by Brain Derived Neurotrophic Factor (BDNF) (Xie &
Yung, 2012), thioredoxin (Nair et al., 2011), or NADPH Oxidase (Guo et al., 2013).
It remains unclear whether such measurements would be of value in individuals with
OSA.
Conclusions
Our findings demonstrate that the relationship of cognitive deficit to sleep
fragmentation and hypoxia in OSA is still unclear. That OSA causes cognitive
Neurocognitive Disturbance in OSA 114
impairment is not in doubt. However, until disease duration can be measured and
specific mechanisms of harm, the nature of protective and risk factors, and their
relationship to specific aspects of cognitive ability are established, there is no basis
on which to identify who is most at risk of cognitive decline, and how to intervene
most effectively.
Neurocognitive Disturbance in OSA 115
CHAPTER 6:
GENERAL DISCUSSION
Overview of findings
This thesis explored the relationship between attention, memory, executive
function, and nocturnal disturbance (hypoxia and sleep fragmentation), in individuals
with OSA.
Chapter 2 (Study 1) used meta-analytic techniques to examine executive
function in individuals with OSA, before and after treatment. It is the first meta-
analysis to divide executive function by its theoretical components and examine the
effects of OSA before and after treatment on these components. The results from this
chapter show that individuals with OSA exhibit widespread impairments across all
facets of executive function, and that these impairments are somewhat ameliorated
with CPAP treatment. It remains to be understood whether the executive
impairments seen are mediated by inter-individual differences in premorbid IQ,
sleepiness, age, and disease severity. Furthermore, it is uncertain if these
impairments represent primary or secondary effects of OSA, that is to say whether
the impairments seen in any one aspect of cognition are direct consequences of OSA
The present chapter was reformatted for submission as an invited paper to
Translational Issues in Psychological Science; Olaithe, M., Bucks, R.S. & Eastwood, P.
Obstructive Sleep Apnoea (OSA), affecting cognition while we’re in the dark:
Implications for practitioners (Submitted 15.04.2014). This chapter has been formatted
for consistency with the rest of the thesis.
Neurocognitive Disturbance in OSA 116
or due to impairment in another domain, such as attention, and whether functioning
returns to baseline with perfectly adherent treatment.
Chapter 4 (Study 2) used Structural Equation Modelling (SEM) to examine the
factor structure of the ESS and to assess the utility of a model with one factor of
sleepiness using all items, versus a model with 3 levels of somnificity. In both
community and sleep clinic samples, a 3-factor structure provided a level of model fit
that was at least as good as the 1-factor structure. These findings suggest that the ESS
can be used to measure an individual’s overall sleep propensity as well as more
specific measures of sleep propensity in low, moderate and high levels of situational
somnificity.
Chapter 5 (Study 3) sought to clarify the relationships between the domains
of cognition that are reported to be impaired in OSA (attention, memory and
executive function) and the proposed mechanisms that cause such impairment (sleep
fragmentation and hypoxia). This study is the first comprehensively to assess the
contributions of the major mechanisms currently thought to be responsible for
cognitive dysfunction in OSA; namely hypoxia and sleep fragmentation (Beebe &
Gozal, 2002; Verstraeten et al., 2004). When controlling for premorbid IQ alone,
more severe sleep fragmentation, but not hypoxia, was significantly associated with
poorer attention, and executive dysfunction, but not with memory dysfunction.
However, when age was accounted for, these effects became non-significant.
Taken together, the studies in this thesis show that OSA has a negative
impact on cognition, but that, after accounting for inter-individual differences in
premorbid IQ and age there is no relationship between the severity of hypoxia and
sleep fragmentation and cognitive impairment.
Neurocognitive Disturbance in OSA 117
Implications of these findings will be discussed with reference to the wider
literature, future directions, and recommendations for practicing psychology
clinicians.
Discussion of findings
OSA is associated with activation in (Zhang et al., 2011) and structural
abnormality of (Zimmerman & Aloia, 2006) areas of the brain associated with
attention, memory and executive function in OSA. OSA also leads to grey matter
loss in frontal (associated with executive function) and temporo-parieto-occipital
cortices (broadly associated with attention and visuospatial processing), the thalamus
(largely involved in relaying information, and regulating sleep-wake cycles) and the
hippocampal region (largely associated with memory) (Yahoui et al., 2009). The
results from Chapter 2 provide further support that OSA leads to dysfunction in the
cognitive facets associated with these areas.
There is additional evidence through examination of chronic diseases with
similar mechanisms of harm (Incalzi et al., 2004), animal models (Row, 2007),
imaging studies (Thomas, Rosen, Stern, Weiss, & Kwong, 2005) and treatment
studies (Ferini-Strambi, Baietto, & Di Gioia, 2003) that sleep fragmentation and
hypoxia negatively impact on the brain and on cognitive functioning. Indeed, the
dominant model of neural harm and cognitive dysfunction accentuates sleep
fragmentation and hypoxia as the main mechanisms of harm in OSA (Beebe &
Gozal, 2002).
However, despite evidence that OSA is associated with cognitive impairment,
and similar to the findings reported in Chapter 5, other papers have not consistently
found a relationship between the degree of nocturnal harm, operationalized by sleep
Neurocognitive Disturbance in OSA 118
fragmentation and hypoxia, and daytime cognitive dysfunction. For example, Aloia
et al. (2004) undertook a review examining evidence of the relationship between
cognitive dysfunction in OSA, sleep fragmentation and hypoxia. This review
concluded that, at present, the evidence is equivocal. More recent reviews focusing
on memory (Wallace & Bucks, 2012), together with the meta-analysis in this thesis,
similarly support the lack of evidence of a relationship between sleep fragmentation,
hypoxia and cognition.
It has previously been suggested that no relationship between has been found
between these due to small sample sizes, use of less than optimal statistical
techniques, and failure to control for inter-individual variables such as cognitive
reserve or sleepiness (Tsai, 2010). These points are discussed with reference to the
present thesis.
One possibility for a lack of consistent association between cognitive
dysfunction, sleep fragmentation and hypoxia is that previous studies have been
underpowered. Examining this line of inquiry, the present thesis used large samples.
Chapter 2 (Study 1) synthesised 24 years of research, reporting combined samples of
551 healthy controls and 1010 individuals with OSA. Chapter 4 (Study 2) obtained
samples of 356 community individuals, and 693 individuals visiting a sleep clinic.
Similarly, Chapter 5 reports cognitive performance in one of the largest samples to
date of individuals with OSA (N = 150). Of the papers reviewed by Beebe et al.
(2003), which examined cognition in OSA, the median sample size was 19 (Range =
7 - 199). Furthermore, the samples in all thesis chapters had a wide range of ages,
and used male and female individuals, from a wide range of occupational settings.
Neurocognitive Disturbance in OSA 119
Therefore, it seems unlikely that a dose-response relationship between hypoxia, sleep
fragmentation and cognition was not found due to lack of power.
Another possibility for a lack of association between cognition and severity of
OSA is that statistical techniques used in previous studies do not account for
measurement error. Many papers have used correlation or traditional regression
which do not allow the modelling of measurement error, nor can multiple measures
of the same hypothesised construct be used to capture cognitive performance (see
Aloia et al., 2004, Table 3, for a summary of papers that found or did not find a
relationship between domains of cognition, sleep fragmentation and/or hypoxia). To
address this concern, the present thesis used Structural Equation Modelling (SEM),
which provides a better fit to the data by accounting for measurement error. The use
of SEM makes it unlikely that a dose-response relationship was not found for
statistical reasons
Last, a number of key inter-individual variables including age and premorbid
intelligence can impact upon the expression of cognitive impairment. These
variables are not routinely accounted for in many studies, and some cannot yet be
quantified, e.g., disease duration. This thesis carefully examined a range of inter-
individual confounds, including age, premorbid intelligence and sleepiness.
Premorbid intelligence was positively related to cognition, which underscores
the importance of controlling for premorbid ability when evaluating cognition in
OSA. This is in line with findings by Alchanatis et al. (2005), who reported that
people with OSA and high intelligence performed equally well to people without
OSA, whilst individuals with OSA and normal intelligence demonstrated cognitive
deficits, compared to controls. Cognitive reserve has been shown to be influential in
Neurocognitive Disturbance in OSA 120
understanding changes in cognitive evidenced in other chronic disorders, including
Alzheimer’s disease (Querbes et al., 2009).
In contrast to premorbid IQ, analysis of subjective sleepiness revealed that
neither ESS total, nor factors of situational somnificity (low, moderate, high
situational somnificity), were related to cognition. The finding that subjectively
measured sleepiness and somnificity are not related to cognition in OSA has been
reported elsewhere (Incalzi et al., 2004). One reason for this might be that
individuals with chronic sleep deprivation, as occurs in OSA, can become
desensitized to the feeling of sleepiness leading to under or over reporting (Durmer
& Dinges, 2005b). In order to take into account such factors, future research could
use objective measures of sleepiness such as the Maintenance of Wakefulness Test
(MWT), and /or the Multiple Sleep Latency Test (MSLT). However, these tests are
not routinely used in research due to their high monetary and time costs. Novel
markers of sleepiness and neurocognitive impairment, for example the Detrended
Fluctuation Analysis (DFA) scaling exponent of the awake electroencephalogram,
may be particularly useful in providing relatively quick and objective measures of
sleepiness (D'Rozario et al., 2013).
This raises the question of why, despite responding to criticisms levelled at
earlier studies regarding sample size, inter-individual differences, or statistical
methods, the present analysis did not find a relationship between sleep fragmentation
and hypoxia and cognitive dysfunction. The results presented here, and those of past
reviews, indicate the very real possibility that there is no relationship to be found.
That is, that OSA causes cognitive impairment but that the degree of such
impairment is a function of age, disease duration, premorbid cognitive reserve and
Neurocognitive Disturbance in OSA 121
reaching some critical threshold of OSA severity above which no additional damage
to cognition is accrued. An alternative explanation is that we need to examine other
possible mechanisms of harm such as hypercapnia, or develop novel metrics that
capture inter-individual differences, and subtle changes in cognition, sleep
fragmentation and hypoxia.
These factors are discussed below under the headings of 1. Incorrect or
imprecise indices of harm and, 2. Incorrect or imprecise measurement of cognition in
OSA and, 3. Additional inter-individual differences.
1. Incorrect or imprecise measurement of nocturnal disturbance
It is possible that sleep fragmentation and hypoxia are not the main
mechanisms of harm in OSA. One poorly investigated but destructive influence on
cerebral tissue is hypercapnia. Indeed, excessive levels of carbon dioxide are found
in OSA (AASM, 2007; Kaw et al., 2009), yet hypercapnia is not routinely measured
in overnight sleep studies nor explored in cognitive research in OSA. What limited
research there is suggests that the degree of hypercapnia correlates with overall
cognitive impairment (Findley et al., 1986).
Alternatively, sleep fragmentation and hypoxia may be the mechanisms of
harm, but routine indices may be insufficiently sensitive to the extent of sleep
disturbance, or the length and depth of hypoxia (Asano et al., 2009). Routine, clinical
sleep fragmentation measures (such as Arousal Index: ArI) capture gross
awakenings. Likewise, routine measures of oxygenation (such as the SpO2:
peripheral capillary oxygen saturation) report only the depth or time spent below a
certain oxygen threshold, or the average number of events per hour. An individual
with OSA may exhibit many shallow but long lasting oxygen desaturations, or
Neurocognitive Disturbance in OSA 122
multiple short complete apnoeas, or they may experience ‘micro-arousals’, a small
fragmentation of sleep architecture, not captured by standard indices. Measures that
capture microarousals may provide more intimate detail of sleep fragmentation
(Pepin et al., 2005), likewise, new measures may describe the length and depth of
desaturation for each individual (e.g. Integrated Area of Desaturation) (Asano et al.,
2009).
2. Incorrect or imprecise measurement of cognition
Research examining the cognitive profile of individuals with OSA has focused
on examining whole domains of cognition, such as ‘attention’, ‘memory’ and
‘executive function’. Such categorisation may simply be too broad.
Neuropsychological tasks typically measure a discrete facet of a whole domain. For
example, a task such as Trails B (the participant is instructed to connect a set of
25 letters and numbers as fast as possible, while maintaining accuracy) is typically
used in OSA research to examine ‘executive function’, however this task most likely
quantifies the capacity of an individual to ‘shift’ their attention between information
sets. This Shifting capacity is only a subset of executive function, and does not
reflect the whole theoretical domain (Fisk & Sharp, 2004; Miyake et al., 2000).
3. Additional inter-individual differences
Recent research suggests that perception of sleep quality affects daytime
functioning and neurocognitive task performance (Draganich & Erdal, 2014). In a
novel experiment Draganich and Erdal (2014) demonstrated that individuals labelled
as having poor sleep quality, regardless of actual sleep quality, performed less well
on a task of attention (Paced Auditory Serial Attention Task, which assesses auditory
information processing speed and flexibility by asking the individual to add up a
Neurocognitive Disturbance in OSA 123
series of numbers) and executive function (Controlled Oral Word Association Task,
which measures verbal fluency by asking individuals to pronounce a set of irregular
verbs) than individuals instructed they had slept well. These authors were able to
show that this was not due to task demands, and instead was solely attributable to
assigned sleep quality. Other researchers have reported that self-perceived sleep
quality can affect cognitive performance (Querbes et al., 2009). Similarly,
individuals with OSA may perceive their sleep quality to be affected by their new
diagnosis of OSA and this may, in turn, affect their performance on
neuropsychological assessment.
The studies in this thesis showed that, having accounted for participant age,
relationships between sleep fragmentation, attention and executive function were no
longer significant. This may be, in part, due to age being confounded with disease
duration. At present the best estimate of disease duration in OSA comes from
subjective reports of how long the individual has exhibited snoring. However, the
individual (and/or partner) may not be aware of their snoring, nor does it
conclusively indicate OSA (Cirignotta et al., 2009). Through examination of other
chronic diseases where hypoxia is a symptom, longer untreated exposure has been
shown to result in greater cumulative damage (Langan, Deary, Hepburn, & Frier,
1991). As such, it is likely that individuals with longer disease duration may exhibit
more severe cognitive dysfunction with less responsiveness to CPAP treatment due
to longer exposure to hypoxia, sleep fragmentation, or some other mechanism of
harm (e.g., hypercapnia). Support for such a notion can be found in studies showing
persistent cognitive difficulties following CPAP therapy (Bardwell, Ancoli-Israel,
Berry, & Dimsdale, 2001; Bedard, Montplaisir, Malo, Richer, & Rouleau, 1993).
Neurocognitive Disturbance in OSA 124
Measurement of disease duration remains a challenge for the field of sleep medicine
and presents a substantial barrier to our capacity to interpret studies related to OSA.
However, it may be possible to develop a method to ascertain disease duration.
In other chronic diseases such as diabetes, long term disease duration can be
measured by the degree of impairment in vascular reactivity of systemic arteries
(Clarkson et al., 1996). Given that endothelial and cerebral tissue damage and loss is
evident in individuals with OSA (Budhiraja, Parthasarathy, & Quan, 2007; Yahoui et
al., 2009), such strategies hold promise for the estimation of disease duration in
OSA. Longitudinal studies would be required to catalogue tissue damage in OSA,
and develop a way of quantifying disease duration with respect to cerebral tissue
damage. Such an examination is already being carried out within the realm of
cardiovascular research (Wang, Wu, Feng, & Sun, 2013).
Implications for clinicians
There is a need for greater awareness of OSA which should assist with earlier
detection of this condition. The negative cognitive and social implications of OSA
are substantial (Al-Ghanim et al., 2008; Hillman et al., 2006a), and to some degree
irreversible (Bedard et al., 1993). Hence, developing systems for early detection is
necessary. Due to the high co-morbidity between depression (Schroder & O'Hara,
2005), cognitive disturbance (Beebe, 2006) and OSA, clinical psychologists and
neuropsychologists are likely to be visited by a high proportion of individuals with
OSA. Furthermore, psychologists receive intimate information about their clients,
and could extend this to asking about sleep quality and habits.
Clinical psychologists and clinical neuropsychologists could potentially be a
well-informed point of contact for individuals with sleep disturbances. Sleep
Neurocognitive Disturbance in OSA 125
physicians are already a well informed source of information about sleep disorders
and treatments, however, due to the overlap of sleep disorders and psychological
disorders (Schroder & O'Hara, 2005), not all sufferers of sleep disorders will be
referred to a sleep physician. There are presently discrete roles for assessing and
addressing sleep disorders and psychological disorders, however due to the overlap
in OSA and psychological disorders, greater communication and collaboration
between fields is required.
Although not all OSA can be picked up by subjective report, or identifying
demographic risk factors, there are a number of features to which clinical
psychologists and clinical neuropsychologists should attend. First, the foremost
demographic risk factors for OSA are older age, obesity, and being male (Young et
al., 1993; T. Young et al., 2002). Second, should an individual report snoring,
frequent nocturnal awakenings with a feeling of dread or gasping for air, feeling un-
refreshed upon awakening, falling asleep or feeling tired during the day, these
individuals are at risk of OSA (Netzer et al., 1999). There are a number of paper-
and-pencil tasks that can be utilised to assess these features. Of particular note is the
Berlin Questionnaire, which is a 10 minute, self-administered questionnaire that
identifies the level of risk that an individual has sleep disordered breathing (Netzer et
al., 1999). The clinician can then recommend the individual visit their general
practitioner, with the Berlin Questionnaire scores.
At present, the psychologist must refer the client to their GP, who can assess
further if a sleep study is warranted; a psychologist cannot directly refer to a sleep
physician. However, this is possibly another point where revision of the health care
system is required.
Neurocognitive Disturbance in OSA 126
Recommending a client has their sleep assessed may assist in the treatment of
the psychological disorder for which the person was first referred. Research
demonstrates that some individuals with co-morbid OSA and depression experience
improvements in depression symptoms with OSA treatment (Schroder & O'Hara,
2005).
In addition, psychologists may have an important role to play to improve
CPAP treatment adherence. Treatment with CPAP improves quality of life, reduces
car accidents and, to some extent, assists with deficits in cognition (Weaver &
Grunstein, 2008). However, between 46-83% of people with OSA are not adherent
to the minimum 4 hours a night of recommended CPAP use (Weaver & Grunstein,
2008).
Illness beliefs may play an important role in this non-adherence. Individuals
who believe that their OSA symptoms or diagnosis is due to a psychological stressor
rather than a physical cause or condition are less likely to adopt a physiological
treatment, such as CPAP (Skinner et al., 2013). Psychologists can assess and address
the impact such beliefs may have on treatment adherence.
Additionally, individuals with little social support demonstrate lower rates of
CPAP adherence (Stepnowsky, Marler, Palau, & Brooks, 2006). Providing social
support or assisting clients to rebuild their social support, may assist with treatment
compliance (Richards, Bartlett, Wong, Malouff, & Grunstein, 2007).
In order to accomplish greater detection and support for individuals with OSA,
psychology clinicians need to be taught about OSA and other sleep disorders.
However, the impact of sleep disturbance on psychological and cognitive health are
not routinely taught in clinical or neuro psychology courses (Waters & Bucks, 2011).
Presently, the Australian Psychological Society (APS) does not list sleep disorders as
Neurocognitive Disturbance in OSA 127
a necessary facet of psychology teaching programs, and most neuropsychological
and clinical training text books do not include information on sleep disorders
(Waters & Bucks, 2011). Despite this, sleep disorders have notable effects on
cognition, psychological health and quality of life (Butkov & Lee-Chiong, 2007). It
is time training institutes routinely teach the impact of, and how to screen for, sleep
disorders in the clinic.
Conclusions
This thesis examined the impact of OSA upon cognition, demonstrating that
OSA impacts upon attention and executive function. As yet, it is unclear which
aspects of the nocturnal disturbance of OSA result in the cognitive dysfunction seen
in OSA, and the precise picture of inter-individual differences that moderate this
harm. It is clear that age and premorbid IQ play a role in moderating cognitive
dysfunction in OSA, however future research is required to identify novel metrics of
nocturnal disturbance, ‘best practice’ ways to measure cognition, and how to capture
disease duration.
Neurocognitive Disturbance in OSA 128
References
AASM. (1999). Sleep–Related Breathing Disorders in Adults: Recommendations for Syndrome Definition and Measurement Techniques in Clinical Research. Sleep, 22, 667-689.
AASM. (2007). The AASM manual for the scoring of sleep and associated events; rules, terminology and technical specifications. Illinois: American Academy of Sleep Medicine.
AASM. (2012). AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications. Illinois: American Academy of Sleep Medicine.
Adrover-Roig, D., Sesé, A., Barceló, F., & Palmer, A. (2012). A latent variable approach to executive control in healthy ageing. Brain and Cognition, 78, 284-299.
Al-Ghanim, N., Comondore, V. R., Fleetham, J., Marra, C. A., & Aya, N. T. (2008). The economic impact of obstructive sleep apnoea. Lung, 186, 7-12.
Al-Lawati, N. M., Patel, S. R., & Ayas, N. T. (2009). Epidemiology, risk factors, and consequences of obstructive sleep apnoea and short sleep duration. Progress in Cardiovascular Disease, 51, 285-293
Alchanatis, Zias, N., Deligiorgis, N., Amfilochiou, A., Dionellis, G., & Orphanidou, D. (2005). Sleep apnea-related cognitive deficits and intelligence: an implication of cognitive reserve theory. Journal of Sleep Research, 14, 69-75.
Alchanatis, Zias, N., Deligiorgis, N., Chroneou, A., Soldatos, C., & Roussos, C. (2008). Comparison of cognitive performance among different age groups in patients with obstructive sleep apnea. Sleep and Breathing, 12(1), 17-24.
Alchanatis, Zias, N., Deligiorgis, N., Liappas, I., Chroneou, A., Soldatos, C., & Roussos, C. (2008). Comparison of cognitive performance among different age groups in patients with obstructive sleep apnea. Sleep and Breathing, 12, 17-24.
Alhola, P., & Polo-Kantola, P. (2007). Sleep deprivation: Impact on cognitive performance. Neuropsychiatric Disease and Treatment, 3(5), 553-567.
Aloia, M. S., Arnedt, T., Davis, J. D., Riggs, R. L., & Byrd, D. (2004). Neuropsychological sequelae of obstructive sleep apnea-hypopnea syndrome: A critical review. Journal of the International Neuropsychological Society, 10, 772-785.
Antic, N. (September, 2012). [Data acquired from the author through email correspondence].
Antic, N. A., Catcheside, P., Buchan, C., Hensley, M., Naughton, M. T., Sharn Rowland, S., . . . McEvoy, R. D. (2011). The Effect of CPAP in Normalizing Daytime Sleepiness, Quality of Life, and Neurocognitive Function in Patients with Moderate to Severe OSA. Sleep, 34, 111-119.
Antonelli-Incalzi, R., Marra, C., Salvigni, B. L., Petrone, A., Gemma, A., Selvaggio, D., & Mormile, F. (2004). Does cognitive dysfunction conform to a distinctive pattern in obstructive sleep apnea syndrome? . Journal of Sleep Research(13), 79-86.
Antony, M. M., Bieling, P. J., Cox, B. J., Enns, M. W., & Swinson, R. P. (1998). Psychometric properties of the 42-item and 21-item versions of the Depression anxiety stress scales in clinical groups and a community sample. Psychological Assessment, 10, 176-181.
Neurocognitive Disturbance in OSA 129
Asano, K., Takata, Y., Usui, Y., Shiina, K., Hashimura, Y., Kato, K., . . . Yamashina, A. (2009). New index for analysis of polysomnography, 'integrated area of desaturation', is associated with high cardiovascular risk in patients with mild to moderate obstructive sleep apnea. Respiration, 78(3), 278-284.
Ayalon, L., Ancoli-Israel, S., & Drummond, S. P. (2010). Obstructive sleep apnea and age: a double insult to brain function? . Am J Respir Crit Care Med, 182, 413-419.
Ayalon, L., Ancoli-Israel, S., Klemfuss, Z., Shalauta, M. D., & Drummond, S. P. A. (2006). Increased brain activation during verbal learning in obstructive sleep apnea. NeuroImage, 31, 1817-1825.
Ayalon, L., Ancoli-lsrael, S., Aka, A. A., McKenna, B. S., & Drummond, S. P. (2009). Relationship between obstructive sleep apnea severity and brain activation during a sustained attention task. Sleep: Journal of Sleep and Sleep Disorders Research, 32(3), 373-381.
Banich, M. T. (2009). Executive Function : The Search for an Integrated Account. Current Directions in Psychological Science, 18, 89-94.
Bardwell, W. A., Ancoli-Israel, S., Berry, C. C., & Dimsdale, J. E. (2001). Neuropsychological Effects of One-Week Continuous Positive Airway Pressure Treatment in Patients With Obstructive Sleep Apnea: A Placebo-Controlled Study. Psychosomatic Medicine, 63, 579-584.
Barnes, M., Houston, D., Worsnop, C. J., Neill, A. M., Mykytyn, I. J., Kay, A., . . . Douglas, M. R. (2002). A Randomized Controlled Trial of Continuous Positive Airway Pressure in Mild Obstructive Sleep Apnea. American Journal of Respiratory Critical Care Medicine, 165, 773-780.
Bartlett, D., Rae, C., Thompson, C., Byth, K., Joffe, D., Enright, T., & Grunstein, R. (2004). Hippocampal area metabolites relate to severity and cognitive function in obstructive sleep apnea. Sleep Medicine, 5(6), 593-596.
Bastien, C. H., Vallieres, A., & Morin, C. M. (1999). Validation of the Insomnia Severity Index as an outcome measure for insomnia research. Sleep Medicine, 2(4), 297-307.
Bedard, M. A., Montplaisir, J., Malo, J., Richer, F., & Rouleau, I. (1993). Persistent neuropsychological deficits and vigilance impairment in sleep apnea syndrome after treatment with continuous positive airways pressure (CPAP). Journal of Clinical and Experimental Neuropsychology, 15(2), 330-341.
Bedard, M. A., Montplaisir, J., Richer, F., Rouleau, I., & Malo, J. (1991). Obstructive sleep apnea syndrome: pathogenesis of neuropsychological deficits. Journal of Clinical and Experimental Neuropsychology, 13(6), 950-964.
Beebe, D. W. (2006). Neurobehavioral effects of obstructive sleep apnea: an overview and heuristic model. Current Opinion in Pulmonary Medicine, 11, 494-500.
Beebe , D. W., & Gozal, D. (2002). Obstructive sleep apnea and the prefrontal cortex: towards a comprehensive model linking nocturnal upper airway obstruction to daytime cognitive and behavioral deficits. Journal of Sleep Research, 11, 1-16.
Beebe, D. W., Groesz, L., Wells, C., Nichols, A., & McGee, K. (2003). The Neuropsychological Effects of Obstructive Sleep Apnea: A Meta-Analysis of Norm-Referenced and Case-Controlled Data. Sleep, 26, 298-307.
Neurocognitive Disturbance in OSA 130
Bloch, K. E., Schoch, O. D., Zhang, J. N., & Russi, E. W. (1999). German version of the Epworth Sleepiness Scale. Respiration, 66, 440-447.
Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2005). Comprehensive Meta-Analysis Version 2 Biostat.
Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2009). Introduction to Meta-Analysis. West Sussex: John Wiley & Sons.
Bucks, R., Olaithe, M., & Eastwood, P. (2012). Neurocognitive function in obstructive sleep apnoea: A meta-review. Respirology, 1-10.
Budhiraja, R., Parthasarathy, S., & Quan, S. F. (2007). Endothelial Dysfunction in Obstructive Sleep Apnea. Journal of Clinical Sleep Medicine, 3(4), 409-415.
Burt, D. B., Zembar, M. J., & Niederehe, G. (1995). Depression and memory impairment: A meta-analysis of the association, its pattern, and specificity. Psychological Bulletin, 117(2), 285-305.
Butkov, N., & Lee-Chiong, T. (2007). Fundamentals of Sleep Technology. Philadelphia: Lippincott Williams & Wilkins.
Buysse, D. J., Reynolds, C. F., & Monk, T. H. (1989). Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiatry Research, 28, 193-213.
Byrne, B. M. (2010). Structural Equation Modelling with AMOS: Basic concepts, application and programming (2nd Ed.). Structural Equation Modelling with AMOS: Basic concepts, application and programming (2nd ed.). New York: Taylor & Francis Group.
Canessa, N., Castronovo, V., Cappa, S. F., Aloia, M. S., Marelli, S., Falini, A., . . . Ferini-Strambi, L. (2011). Obstructive Sleep Apnea: Brain Structural Changes and Neurocognitive Function before and after Treatment. American Journal of Respiratory and Critical Care Medicine, 183, 1419-1426.
Carlson, B. W., Neelon, V. J., Carlson, J. R., Hartman, M., & Bliwise, D. L. (2011). Cerebral Oxygenation in Wake and During Sleep and Its Relationship to Cognitive Function in Community-Dwelling Older Adults Without Sleep Disordered Breathing. Journal of Gerontology, 66A(1), 150-156.
Carskadon, M. A., & Dement, W. C. (2011). Monitoring and staging human sleep. In M. H. Kryger, T. Roth & W. C. Dement (Eds.), Principles and Practice of Sleep Medicine (5th ed.). St. Louis: Elsevier Saunders.
Casseli, R. J. (2008). Obstructive sleep apnea, apolipoprotein E e4, and mild cognitive impairment. Sleep Medicine, 9, 816-817.
Castronovo, V., Canessa, N., Ferini-Strambi, L., Aloia, M. S., Consonni, M., Marelli, S., . . . Cappa, S. F. (2009). Brain Activation Changes Before and After PAP Treatment in Obstructive Sleep Apnea. Sleep, 32(9), 1161-1172.
Chan, R. C. K., Chen, E. Y. H., Cheung, E. F. C., Chen, R. Y. L., & Cheung, H. K. (2006). The components of executive functioning in a cohort of patients with chronic schizophrenia: A multiple single-case study design. Schizophrenia Research, 81, 173-189.
Chen, N. H., Johns, M. W., Li, H. Y., Chu, C. C., Liang, S. C., Shu, Y. H., . . . Wang, P. C. (2002). Validation of a Chinese version of the Epworth sleepiness scale. Quality of Life Research, 11(8), 817-821.
Cheng, H. C., Dai, Z. K., Wu, J. R., & Chen, I. C. (2012). Pediatric Upper Airway Emergencies. Taiwan Society of Pediatric Pulmonology, 25-31.
Chiang, A. A. (2006). Obstructive sleep apnea and chronic intermittent hypoxia: a review. Chinese Journal of Physiology, 49(5), 234-243.
Neurocognitive Disturbance in OSA 131
Cirignotta, F., D'Alessandro, R., Partinen, M., Zucconi, M., Cristina, E., Gerardi, R., . . . Lugaresi, E. (2009). Prevalence of every night snoring and obstructive sleep apnoeas among 30-69-year-old men in Bologna, Italy. 79, 5(366-72).
Clarkson, P., Celermajer, D. S., Donald, A. E., Sampson, M., Sorensen, K. E., Adams, M., . . . Deanfield, J. E. (1996). Impaired vascular reactivity in insulin-dependant diabetes mellitus is related to disease duration and low density lipoprotein cholesterol levels. JACC. Cardiovascular Imaging, 28(3), 573-579.
Cohen, J. (1977). Statistical power analysis for the behavioral sciences. New York: Academic Press.
Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.
Corporation, U. B. (2009). Computerised Drug Research System. London: Bracket. Cowan, D. C., & Livingston, E. (2012). Obstructive Sleep Apnoea Syndrome
andWeight Loss: Review. Sleep Disorders, 1-11. Crawford, J. R., Deary, I. J., Starr, J., & Whalley, L. J. (2001). The NART as an
index of prior intellectual functioning: a retrospective validity study covering a 66-year interval. Psychological Medicine, 31, 451-458.
Crawford, J. R., Stewart, L. E., Cochrane, R. H. B., Parker, D. M., & Besson, J. A. O. (1989). Construct validity of the National Adult Reading Test: A factor analytic study. Personality and Individual Differences, 10, 585-587.
D'Rozario, A. L., Kim, J. W., Wong, K. K. H., Bartlett, D. J., Marshall, N. S., Dijk, D. J., . . . Grunstein, R. R. (2013). A new EEG biomarker of neurobehavioural impairment and sleepiness in sleep apnea patients and controls during extended wakefulness. Clinical Neurophysiology, 124(8), 1605-1614.
Data, S., & MacLean, R. R. (2007). Neurobiological mechanisms for the regulation of mammalian sleep–wake behavior: Reinterpretation of historical evidence and inclusion of contemporary cellular and molecular evidence. Neuroscience and Behavioral Reviews, 31(5).
Daurat, A., Foret, J., Bret-Dibat, J. L., Fureix, C., & Tiberge, M. (2008). Spatial and temporal memories are affected by sleep fragmentation in obstructive sleep apnea syndrome. Journal of Clinical & Experimental Neuropsychology, 30(1), 91-101.
Deegan, J. J. (1978). On the Occurrence of Standardized Regression Coefficients Greater Than One. Educational and Psychological Measurement, 38(4), 873-888.
Draganich, C., & Erdal, K. (2014). Placebo sleep affects cognitive functioning. Journal of Experimental Psychology: Learning, Memory, and Cognition, EPub ahead of print.
Durmer, J. S., & Dinges, D. F. (2005a). Neurocognitive Consequences of Sleep Deprivation. Seminars in Neurology, 25(1), 117-129.
Durmer, J. S., & Dinges, D. F. (2005b). Neurocognitive Consequences of Sleep Deprivation. Seminars in Neurology, 25(117-129).
Duval, S., & Tweedie, R. (2000). Trim and fill: A simple funnel-plot based method of testing adn adjusting for publication bias in meta-analysis. Biometrics, 56, 455-463.
Egger, M., Smith, G. D., Schneider, M., & Minder, C. (1997). Bias in meta-analysis detected by a simple, graphical test. British Medical Journal, 315, 629-634.
Neurocognitive Disturbance in OSA 132
Elliot, R. (2003). Executive functions and their disorders. British Medical Bulletin, 65, 49–59.
Engleman, H. M., & Douglas, N. J. (2004). Sleep 4: Sleepiness, cognitive function, and quality of life in obstructive sleep apnoea/hypopnoea syndrome. Thorax, 59, 618-622.
Estrada, A., Kelley, A. M., Webb, C. M., Athy, J., & Crowley, J. S. (2012). Modafinil as a Replacement for Dextroamphetamine for Sustaining Alertness in Military Helicopter Pilots Aviation, Space, and Environmental Medicine, 83, 556-567.
Everson, C. A. (1995). Functional consequences of sustained sleep deprivation in the rat. Behavioural Brain Research, 69, 43-54.
Ferini-Strambi, L., Baietto, C., & Di Gioia, M. R. (2003). Cognitive dysfunction in patients with obstructive sleep apnea (OSA): partial reversibility after continuous positive airway pressure (CPAP). Brain Research Bulletin, 61, 87-92.
Findley, L. J., Barth, J. T., Powers, D. C., Wilhoit, S. C., Boyd, D. G., & Suratt, P. M. (1986). Cognitive impairment in patients with obstructive sleep apnea and associated hypoxemia. Chest, 90(5), 686.
Fisk, J. E., & Sharp, C. A. (2004). Age-Related Impairment in Executive Functioning: Updating, Inhibition, Shifting, and Access. Journal of Clinical and Experimental Neuropsychology, 26, 874-890.
Floyd, F. J., & Widaman, K. F. (1995). Factor analysis in the development and refinement of clinical assessment instruments. Psychological Assessment, 7, 286-299.
Fulda, S., & Schulz, H. (2001). Cognitive dysfunction in sleep disorders. Sleep Medicine Reviews, 5, 243-445.
Fulda, S., & Schulz, H. (2003). Cognitive dysfunction in sleep-related breathingdisorders: a meta-analysis. Sleep Research Online, 5, 19-51.
Funahashi, S. (2001). Neuronal mechanisms of executive control by the prefrontal cortex. Neuroscience Research, 39, 147–165.
Gale, S. D., & Hopkins, R. O. (2004). Neuroimaging and neuropsychological findings following carbon monoxide poisoning and obstructive sleep apnea. Journal of the International Neuropsychological Society, 10, 60-71.
Green, S. B., Lissitz, R. W., & Mulaik, S. W. (1977). Limitations of coefficient alpha as an index of test unidimensionality. Educational and Psychological Measurement, 37, 827-838.
Guilleminault, C., & Brooks, S. N. (2001). Excessive daytime sleepiness; a challenge for the practising neurologist. Brain, 124, 1482-1491.
Guo, Q., Yan, W., Qing, L., Min, L., & Huan, W. (2013). Levels of thioredoxin are related to the severity of obstructive sleep apnea: based on oxidative stress concept. Sleep and Breathing, 17, 311-316.
Guzman-Marin, R., Bashir, T., Suntsova, N., Szymusiak, R., & McGinty, D. (2007). Hippocampal neurogenesis is reduced by sleep fragmentation in the adult rat. Neuroscience, 148, 325-333.
Hamilton, G. S., & Naughton, M. T. (2013). Impact of obstructive sleep apnoea on diabetes and cardiovascular disease. The Medical Journal of Australia, 199(8), S27-30.
Hatipoglu, U., & Rubinstein, I. (2004). Inflammation and obstructive sleep apnea syndrome: how many ways do I look at thee? Chest editorials, 126, 1-2.
Neurocognitive Disturbance in OSA 133
Heaton, K., & Anderson, D. (2007). A psychometric evaluation of the Epworth Sleepiness Scale. Journal of Nursing Measurement, 15, 177-188.
Henry, J. D., & Crawford, J. R. (2005). The short-form version of the Depression Anxiety Stress Scales (DASS-21): Construct validity and normative data in a large non-clinical sample. British Journal of Clinical Psychology, 44, 227-239.
Hillman, D. R., Murphy, A. S., Antic, R., & Pezzulo, L. (2006a). The economic cost of sleep disorders. Sleep, 29(3), 299-305.
Hillman, D. R., Murphy, A. S., Antic, R., & Pezzulo, L. P. (2006b). The economic cost of sleep disorders. Sleep, 29, 299-305.
Hirshkowitz, M., Black, J. E., Wesnes, K. A., Niebler, G., Arora, S., & Roth, T. (2007). Adjunct armodafinil improves wakefulness and memory in obstructive sleep apnea/hypopnea syndrome. Respiratory Medicine, 101, 616-627.
Hopkins, R. O., Kesner, R. P., & Goldstein, M. (1995). Memory for novel and familiar spatial and linguistic temporal distance information in hypoxic subjects. Journal of the International Neuropsychological Society, 1, 454-468.
Hopkins, R. O., & Woon, F. L. M. (2006). Neuroimaging, Cognitive, and Neurobehavioral Outcomes Following Carbon Monoxide Poisoning. Behavioural and Cognitive Neuroscience Reviews, 5(3), 141-155.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Struct. Equ. Model, 6, 1-55.
Hultsch, D. F., Hertzog, C., Small, B. J., & Dixon, R. A. (1999). Use it or lose it: Engaged lifestyle as a buffer of cognitive decline in aging? Psychology and Ageing, 14(2), 245-263.
Iaccobucci, D. (2010). Structural equation modelling: fit indices, sample size, and advanced topics. Journal of Consumer Psychology, 20, 90-98.
IBM. (2012a). IBM AMOS Statistics for Windows (Version 21.0). Armonk: IBM Corporation.
IBM. (2012b). IBM SPSS Statistics for Windows (Version 21.0). Armonk: IBM Corporation.
Incalzi, R. A., Marra, C., Salvigni, B. L., Petrone, A., Gemma, A., Selvaggio, D., & Mormile, F. (2004). Does cognitive dysfunction conform to a distinctive pattern in obstructive sleep apnea syndrome? Journal of Sleep Research, 13, 79-86.
Izci, B., Ardic, S., Firat, H., Sahin, A., Altinors, M., & Karacan, I. (2008). Reliability and validity studies of the Turkish version of the Epworth Sleepiness Scale. Sleep and Breathing, 12(161-168).
Johns, M. W. (1991). A new method for measuring daytime sleepiness: The Epworth sleepiness scale. Sleep, 14, 540-545.
Johns, M. W. (1992). Reliability and factor-analysis of the Epworth Sleepiness Scale. Sleep, 15, 376-381.
Johns, M. W. (1994). Sleepiness in different situations measured by the Epworth Sleepiness Scale Sleep, 17, 703-710.
Johns, M. W. (2002). Sleep propensity varies with behaviour and the situation in which it is measured: The concept of somnificity. J. of Sleep Res, 11, 61-67.
Neurocognitive Disturbance in OSA 134
Johns, M. W., & Hocking, B. (1997). Daytime sleepiness and sleep habits of Australian workers. Sleep, 10, 844-849.
Joreskog, K. G. (1999). How large can a standardized coefficient be. SSI Central, Inc. . Unpublished report.
Kales, A., Caldwell, A. B., Cadieux, R. J., Vela-Bueno, A., Ruch, L. G., & Mayes, S. D. (1985). Severe obstructive sleep apnea—II: Associated psychopathology and psychosocial consequences. Journal of Chronic Diseases, 38, 427-434.
Kato, M., Roberts-Thomson, P., Phillips, B. G., Haynes, W. G., Winicki, M., Accurso, V., & Somers, V. K. (2000). Impairment of endothelium-dependant vasodilation of resistance vessels in patients with obstructive sleep apnoea. Circulation Journal, 102, 2607-2610.
Kaw, R., Hernandez, A. V., Walker, E., Aboussouan, L., & Mohlesi, B. (2009). Determinants of Hypercapnia in Obese Patients With Obstructive Sleep Apnea: A Systematic Review and Metaanalysis of Cohort Studies. Chest, 136(3), 787-796.
Kingshott, R. N., Engleman, H. M., Deary, I. J., & Douglas, N. J. (1998). Does arousal frequency predict daytime function? European Respiratory Journal, 12, 1264-1270.
Kiralti, P. O., Demir, A. U., Volkan-Salanci, B., Demir, B., & Sahin, A. (2010). Cerebral blood flow and cognitive function in obstructive sleep apnea syndrome Hellenic Journal of Nuclear Medicine, 13(2), 138-143.
Kortte, K. B., Horner, M. D., & Windham, W. K. (2002). The Trail Making Test, Part B: Cognitive Flexibility or Ability to Maintain Set? Applied Neuropsychology, 9(2), 106-109.
Kramar, A. F., Erikson, K. I., & Colcombe, J. (2006). Exercise, cognition, and the aging brain. Journal of Applied Physiology, 101, 1237-1242.
Kribbs, N. B., Pack, A. I., Kline, L. R., Getsy, J. E., Schuett, J. S., Henry, J. N., . . . Dinges, D. F. (1993). Effects of One Night without Nasal CPAP Treatment on Sleep and Sleepiness in Patients with Obstructive Sleep Apnea. American Reviews of Respiratory Disorders, 147, 1162-1168.
Kryger, M. H. (2010). Atlas of Clinical Sleep Medicine. Philadelphia: Saunders Elsevier.
Lamberty, G. J., Putnam, S. H., Chatel, D. M., Bieliauskas, L. A., & Adams, K. M. (1994). Derived Trail Making Test indices: A preliminary report. Neuropsychiatry, Neuropsychology, and Behavioral Neurology, 7, 230-234.
Langan, S. J., Deary, I. J., Hepburn, D. A., & Frier, B. M. (1991). Cumulative cognitive impairment following recurrent severe hypoglycaemia in adult patients with insulin-treated diabetes mellitus. Diabetologia, 34(5), 337-344.
Larner, A. J. (2008). Neuropsychological Neurology; The neurocognitive impairments of neurological disorders. Cambridge: Cambridge University Press.
LaRue, A. (2010). Healthy brain aging: role of cognitive reserve, cognitive stimulation, and cognitive exercises. Clinical Geriatric Medicine, 26, 99-111.
Lehto, J. E., Juuja¨rvi, P., Kooistra, L., & Pulkkinen, L. (2003). Dimensions of executive functioning:Evidence from children. British Journal of Developmental Psychology, 21, 59-80.
Lei, M., & Lomax, R. G. (2005). The Effect of Varying Degrees of Nonnormality in Structural Equation Modeling. Struct. Equ. Model, 12(1), 1-27.
Neurocognitive Disturbance in OSA 135
Lezak, M. D., Howieson, D. B., & Loring, D. W. (2004). Neuropsychological Assessment (4th ed.). New York: Oxford University Press.
Lim, D. C., & Pack, A. I. (2013). Obstructive sleep apnea and cognitive impairment: Addressing the blood-brain barrier. Sleep Medicine Reviews, In Press.
Lin, H., Chan, R. C. K., Zheng, L., Yang, T., & Wang, Y. (2007). Executive functioning in healthy elderly Chinese people. Archives of Clinical Neuropsychology, 22, 501–511.
Lovibond, S. H., & Lovibond, P. F. (1995). Manual for the Depression Anxiety Stress Scales (2nd ed.). Sydney: Psychology Foundation.
Lyznicki, J. M., Doege, T. C., Davis, R. M., & Williams, M. A. (1998). Sleepiness, driving and motor vehicle crashes. Journal of the American Medical Association, 279, 1908-1913.
MacCallum, R. C., & Austin, J. T. (2010). Applications of structural equation modelling in psychology research. Annual Review of Psychology, 51, 201-226.
Macey, P. M., Macey, K. E., Henderson, L. A., Alger, J. R., Frysinger, R. C., Woo, M. A., . . . Harper, R. M. (2003). Functional magnetic resonance imaging responses to expiratory loading in obstructive sleep apnea. Respiratory Physiology and Neurobiology, 38, 275-290.
Maes, J. H. R., Eling, P. A. T. M., Reelick, M. F., & Kessels, R. P. C. (2011). Assessing executive functioning: On the validity, reliability, and sensitivity of a click/point random number generation task in healthy adults and patients with cognitive decline. Journal of Clinical and Experimental Neuropsychology, 33, 366-378.
Marcus, J. A., Pothineni, A., Marcus, C. Z., & Bisognano, J. D. (2014). The Role of Obesity and Obstructive Sleep Apnea in the Pathogenesis and Treatment of Resistant Hypertension. Current Hypertension Reports, 16(411).
Marshall, Dawson, A. D. G., & Bucks, R. S. (2009). Killer in the Cab? Journal of Sleep and Biological Rhythms, 7(Suppl. 1), A54. P066.
McGurn, B., Starr, J. M., Topfer, J. A., Pattie, A., Whiteman, M. C., Lemmon, H. A., . . . Deary, I. J. (2004). Pronunciation of irregular words is preserved in dementia, validating premorbid IQ estimation. Neurology, 62(7), 1184-1186.
McKenna, J. T., Tartar, J. L., Ward, C. P., Thakkar, M. M., Cordeira, J. W., McCarley, R. W., & Strecker, R. E. (2007). Sleep fragmentation elevates behavioural, electrographic and neurochemical measures of sleepiness. Neuroscience, 146, 1462-1473.
Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., & Howerter, A. (2000). The Unity and Diversity of Executive Functions and Their Contributions to Complex ‘‘Frontal Lobe’’ Tasks: A Latent Variable Analysis. Cognitive Psychology, 41, 49-100.
Montgomery, C., Fisk, J. E., Newcombe, R., & Murphy, P. N. (2005). The differential effects of ecstasy/polydrug use on executive components: shifting, inhibition, updating and access to semantic memory. Psychopharmacology, 182, 262–276.
Morin, C. M., & Azrin, N. H. (1985). Social and clinical validation of insomnia treatment outcome. Paper presented at the meeting of Association for Advancement of Behavior Therapy
Neurocognitive Disturbance in OSA 136
Mulgrew, A. T., Ryan, C. F., Fleetham, J. A., Cheema, R., Fox, N., Koehoorn, M., . . . Ayas, N. T. (2007). The impact of obstructive sleep apnea and daytime sleepiness on work limitation. Sleep Medicine, 9, 42-53.
Nair, D., Dayyat, E. A., Zhang, S. X., Wang, Y., & Gozal, D. (2011). Intermittent Hypoxia-Induced Cognitive Deficits Are Mediated by NADPH Oxidase Activity in a Murine Model of Sleep Apnea. PLoS ONE, 6(5), Epub Ahead of Publication.
Naismith, S., Winter, V., Gotsopoulos, H., Hickie, I., & Cistulli, P. (2004). Neurobehavioral Functioning in Obstructive Sleep Apnea: Differential Effects of Sleep Quality,Hypoxemia and Subjective Sleepiness. Journal of Clinical & Experimental Neuropsychology, 26(1), 43-54.
Naismith, S., Winter, V., Gotsopoulos, H., Hickie, I., & Cistulli, P. (2011). Neurobehavioral Functioning in Obstructive Sleep Apnea: Differential Effects of Sleep Quality, Hypoxemia and Subjective Sleepiness. Journal of Clinical and Experimental Neuropsychology, 26(1), 43-54.
Nelson, H. E., & Willison, J. (1991). National Adult Reading Test Manual (2nd ed.). Wilson: NFER-Nelson.
Netzer, N. C., Stoohs, R. A., Netzer, C. M., Clark, K., & Strohl, K. P. (1999). Using the Berlin questionnaire to identify patients at risk for the Sleep Apnoea Syndrome. Annals of Internal Medicine, 131(7), 485-491.
Newhouse, P., Kellar, K., Aisen, P., White, H., Wesnes, K. A., Coderre, E., . . . Levin, E. D. (2012). Nicotine Treatment of Mild Cognitive Impairment: A Six Month Double Blind Pilot Clinical Trial. Neurology, 84, 91-101.
Norbury, R., Craig, M., Cutter, W. J., Whitehead, M., & Murphy, D. (2004). Oestrogen: brain ageing, cognition and neuropsychiatric disorder. Menopause Int., 10(118-22).
Norton, P. J. (2007). Depression Anxiety and Stress Scales (DASS-21): psychometric analysis across four racial groups. Anxiety, Stress and Coping, 20(3), 253-265.
Olaithe, M., & Bucks, R. (2013). Executive Dysfunction in OSA Before and After Treatment: A Meta-Analysis. Sleep, 36(9), 1297-1305.
Olaithe, M., Skinner, T. C., Clarke, J., Eastwood, P., & Bucks, R. S. (2012). Can we get more from the Epworth Sleepiness Scale than just one score?: A confirmatory factor analysis of the ESS. Sleep and Breathing, 17(2), 763-769.
Ong, C. W., O’Driscoll, D. M., Truby, H., Naughton, M. T., & Hamilton, G. S. (2012). The reciprocal interaction between obesity and obstructive sleep apnoea. Sleep Medicine Reviews, 1-9.
Partington, J. E., & Leiter, R. G. (1949). Partington's Pathway Test. The Psychological Services Center Bulletin, 1, 9-20.
Pearson, V. E., Allen, R. P., Dean, T., Gamaldo, C. E., Lesage, S. R., & Earley, C. J. (2006). Cognitive deficits associated with restless legs syndrome (RLS). Sleep Medicine, 7(1), 25-30.
Pepin, J. L., Delavie, N., Pin, I., Deschaux, C., Argod, J., Bost, M., & Levy, P. (2005). Pulse Transit Time Improves Detection of Sleep Respiratory Events and Microarousals in Children. Chest, 127, 722-730.
Perlis, M. L., Jungquist, C., Smith, M., & Posner, D. (2005). Cognitive behavioural treatment of insomnia: A session-by-session guide. . New York: Springer Science+Business Media, LLC.
Neurocognitive Disturbance in OSA 137
Phillips, B. A., Schmitt, F. A., Berry, D. T., Lamb, D. G., Amin, M., & Cook, Y. R. (1990). Treatment of obstructive sleep apnea. A preliminary report comparing nasal CPAP to nasal oxygen in patients with mild OSA. Chest, 98, 325-330.
Poe, G., Walsh, C. M., & Bjorness, T. E. (2010). Cognitive neuroscience of sleep. Progress in Brain Research, 185, 1-19.
Quan, S. F., Wright, R., Baldwin, C. M., Kaemingk, K. L., Goodwin, J. L., Kuop, T. F., . . . Bootzin, R. R. (2006). Obstructive sleep apnea–hypopnea and neurocognitive functioning in the Sleep Heart Health Study. Sleep Medicine, 7, 498-507.
Querbes, O., Aubry, F., Pariente, J., Lotterie, J. A., Demonet, J. F., Duret, V., . . . Celsis, P. (2009). Early diagnosis of Alzheimer's disease using cortical thickness: impact of cognitive reserve. Brain, 132, 2036-2047.
Ray, K., Chatterjee, A., Panjwani, U., Kumar, S., Sahu, S., Ghosh, S., . . . Anand, J. P. (2012). Modafinil improves event related potentials P300 and contingent negative variation after 24 h sleep deprivation. Life Sciences(91), 94-99.
Razi, E., & Akbari, H. (2006). A Comparison of Arterial Oxygen Saturation Measured Both by Pulse Oximeter and Arterial Blood Gas Analyzer in Hypoxemic and Non-hypoxemic Pulmonary Diseases. Turkish Repiratory Journal, 7(2), 43-47.
Rechtschaffen, A., & Kales, A. (1968). A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. California: Brain Information Services/Brain Research Institute - Neurological Information Network.
Redline, S., Strauss, M. E., Adams, N., Winters, M., Roebuck, T., Spry, K., . . . Adams, K. (1997). Neuropsychological Function in Mild Sleep-Disordered Breathing. Sleep, 20, 160-167.
Regard, M., Oelz, O., Brugger, P., & Landis, M. (1989). Persistent cognitive impairment in climbers after repeated exposure to extreme altitude. Neurology, 39(2), 210-213.
Reitan, R. M. (1958). Validity of the trail making test as an indicator of organic brain damage. Perceptual and Motor Skills, 8, 271-276.
Richards, D., Bartlett, D. J., Wong, K., Malouff, J., & Grunstein, R. R. (2007). Increased adherence to CPAP with a group cognitive behavioral treatment intervention: a randomized trial. Sleep, 30(5), 635-640.
Rosenthal, R. (1979). The "file drawer problem" and tolerance for null results. Psychological Bulletin
86, 185-193. Rosenthal, R. (1995). Writing Meta-Analytic Reviews. Psychological Bulletin, 118,
183-192. Roth, T., Rippon, G. A., & Arora, S. (2008). Armodafinil improves wakefulness and
long term episodic memory in nCPAP adherent patients with excessive sleepiness associated with obstructive sleep apnea. Sleep and Breathing, 12, 53-62.
Roth, T., White, D., Schmidt-Nowara, W., Wesnes, K. A., Niebler, G., Arora, S., & Roth, T. (2006). Effects of armodafinil in the treatment ofresidual excessive sleepiness associated with obstructive sleep apnea/hypopnea syndrome: a 12 week,multicenter, double blind,randomized, placebo controlled study in nCPAP adherent adults. Clinical Therapeutics, 28, 689-706.
Neurocognitive Disturbance in OSA 138
Row, B. W. (2007). Intermittent hypoxia and cognitive function: implications from chronic animal models. Hypoxia and Circulation: Advances in experimental medicine and biology, 618, 51-68.
Royall, D. R., Cordes, J. A., & Polk, M. (1998). CLOX: an executive clockdraw-ing task. Journal of Neurology, Neurosurgery and Psychiatry, 64, 588-594.
Sassani, A., Findley, L. J., Kryger, M. D., Goldlust, E., George, C., & Davidson, T. M. (2004). Reducing motor-vehicle collisions, costs, and fatalities by treating obstructive sleep apnoea. Sleep, 27, 453-458.
Saunamäki, T., Himanen, S.-L., Polo, O., & Jehkonen, M. (2010). Executive Dysfunction and Learning Effect after Continuous Positive Airway Pressure Treatment in Patients with Obstructive Sleep Apnea Syndrome. European Neurology, 63, 215-220.
Saunamäki, T., & Jehkonen, M. (2007). A review of executive functions in obstructive sleep apnea syndrome. Acta Neurol Scand., 115, 1-11.
Sauter, C., Senbaum, S. A., Popvic, R., Bauer, H., Lamm, C., Klosch, G., & Zeitlhofer, J. (2010). Excessive daytime sleepiness in patients suffering from different levels of obstructive sleep apnoea syndrome. Journal of Sleep Research, 9(293-301).
Schreiber, J. B., Stage, F. K., King, J., Nora, A., & Barlow, E. A. (2006). Reporting Structural Equation Modeling and Confirmatory Factor Analysis Results: A Review. The Journal of Educational Research, 99(6), 323-337.
Schretlen, D., Buffington, A., Meyer, S., & Pearlson, G. (2005). The use of word-reading to estimate “premorbid” ability in cognitive domains other than intelligence. Journal of the International Neuropsychological Society, 1, 784-787.
Schroder, C. M., & O'Hara, R. (2005). Depression and obstructive sleep apnoea (OSA). Annals of General Psychiatry, 4(13), 1-8.
Shahid, A., Shen, J., & Shapiro, C. (2010). Measurements of sleepiness and fatigue. Journal of Psychometric Research, 69, 81-89.
Sharwood, L. N., Elkington, J., Stevenson, M., Grunstein, R. R., Meuleners, L., Ivers, R. Q., . . . Wong, K. K. (2012). Assessing Sleepiness and Sleep Disorders in Australian Long-Distance Commercial Vehicle Drivers: Self-Report Versus an “At Home” Monitoring Device. Sleep, 35, 469-475.
Skinner, T. C., McNeil, L., Olaithe, M., Eastwood, P., Hillman, D. R., Phang, J., . . . Bucks, R. S. (2013). Predicting uptake of Continuous Positive Airway Pressure (CPAP) therapy in Obstructive Sleep Apnoea (OSA): a belief-based theoretical approach. Sleep and Breathing, 17, 1229-1240.
Smith, S. S., Oei, T. P. S., Douglas, J. A., Brown, I., Jorgensen, G., & Andrews, J. (2008). Confirmatory factor analysis of the Epworth Sleepiness Scale (ESS) in patients with obstructive sleep apnoea. JOurnal of Sleep Medicine, 9(739-44).
Stepanski, E. J. (2002). The Effect of Sleep Fragmentation on Daytime Function. Sleep, 25(2), 268-276.
Stepnowsky, C. J., Marler, M. R., Palau, J., & Brooks, J. A. (2006). Social-cognitive correlates of CPAP adherence in experienced users. Sleep Medicine, 7(4), 350-356.
Stern, Y. (2002). What is cognitive reserve? Theory and research application of the reserve concept. Journal of the International Neuropsychological Society, 8, 448-460.
Neurocognitive Disturbance in OSA 139
Strauss, E., Sherman, E. M. S., & Spreen, O. (2006). A Compendium of Neuropsychological Tests: Administration, Norms, and Commentary (3rd ed.). Oxford: Oxford University Press.
Suchy, Y. (2009). Executive Functioning: Overview, Assessment, and Research Issues for Non-Neuropsychologists. ann. behav. med., 37, 106-116.
Suleiman, K. H., & Yates, B. C. (2011). Translating the Insomnia Severity Index into Arabic. Journal of Nursing Scholarship, 43(1), 49-53.
Sulit, L., Siorfer-Isser, A., Kirchner, L., & Redline, S. (2006). Differences in Polysomnography Predictors for Hypertension and Impaired Glucose Tolerance. Sleep, 29(6), 777-783.
Sullivan, C. E., Berthon-Jones, M., Issa, F. G., & Eves, L. (1981). Reversal of obstructive sleep apnea by continuous positive airway pressure applied through the nares. Lancet, 8225, 862.
Sullivan, C. E., & Issa, F. G. (1985). Obstructive sleep apnea. Philadelphia: WB Saunders.
Szabó, M. (2010). The short version of the Depression Anxiety Stress Scales (DASS-21): Factor structure in a young adolescent sample. Journal of Adolescence, 33(1), 1-8.
Takegami, M., Morita, S., Chin, K., Akashiba, T., Kimura, H., Suzukamo, Y., & Fukuhara, S. (2011). Impact of Excessive Daytime Sleepiness on Health-Related Quality of Life in Patients with Obstructive Sleep Apnea Syndrome. Quality of Life Research, 13(1555).
Tartar, J. L., McKenna, J. T., Ward, C. P., McCarley, R. W., Strecker, R. E., & Brown, R. E. (2010). Sleep fragmentation reduces hippocampal CA1 pyramidal cell excitability and response to adenosine. Neuroscience Letters, 469, 1-5.
Thomas, R. J., Rosen, B. R., Stern, C. E., Weiss, J. W., & Kwong, K. K. (2005). Functional imaging of working memory in obstructive sleep-disordered breathing. Journal of Applied Physiology, 98, 226-234.
Tombaugh, T. N., Kozk, J., & Rees, L. (1999). Normative data stratified by age and education for two measures of verbal fluency: FAS and animal naming. Archives of Clinical Neuropsychology, 14(2), 167-177.
Torun-Yazihan, N., Aydin, H., & Karakas, S. (2007). Neuropsychological profiles in levels of obstructive sleep apnea-hypopnea syndrome. Sleep and Biological Rhythms, 5, 85-94.
Tsai, J. C. G. (2010). Neurological and neurobehavioural sequelae of obstructive sleep apnea. Neurorehabilitation, 26, 85-94.
Ullman, J. B., & Bentler, P. M. (2012). Structural Equation Modeling. In I. B. Weiner (Ed.), Handbook of psychology (2nd ed.). Wiley Online Library: John Wiley & Sons.
Valencia-Flores, M., Bliwise, D. L., Guilleminault, C., Cilveti, R., & Clerk, A. (1996). Cognitive function in patients with sleep apnea after acute nocturnal nasal continuous positive airway pressure (CPAP) treatment: sleepiness and hypoxemia effects. . Journal of Clinical & Experimental Neuropsychology: Official Journal of the International Neuropsychological Society(18), 197-210.
Van Den Goor, J. M., Saxby, B. K., Tijssen, T. G., Wesnes, K. A., de Mol, B. A., & Nieuwland, R. (2008). Improvement of cognitive test performance in patients
Neurocognitive Disturbance in OSA 140
undergoing primary CABG and other CPB-assisted cardiac procedures. Perfusion, 23, 267-273.
Vasudev, A., Saxby, B. K., Colloby, S. O., Brien, J. T., Firbank, M. J., Brooker, H., . . . Thomas, A. J. (2012). Relationship between cognition,magnetic resonance whitematter hyperintensities and cardiovascular autonomic changes in late life depression. American Journal of Geriatric Psychiatry, 20(8), 691-699.
Verstraeten, E. (2007). Neurocognitive effects of obstructive sleep apnea syndrome. Current Neurology and Neuroscience Reports, 7, 161-166.
Verstraeten, E., & Cluydts, R. (2004). Executive control of attention in sleep apnea patients: theoretical concepts and methodological considerations. Sleep Medicine Reviews, 8(4), 257-267.
Verstraeten, E., Cluydts, R., Pevemagie, D., & Hoffman, G. (2004). Executive Function in Sleep Apnea: Controlling for Attentional Capacity in Assessing Executive Attention. Sleep, 27(4), 685-693.
Wallace, A., & Bucks, R. S. (2012). Memory and Obstructive Sleep Apnoea: A Meta-Analysis. Sleep.
Wang, Q., Wu, Q., Feng, J., & Sun, X. (2013). Obstructive sleep apnea and endothelial progenitor cells. Patient Prefer Adherence, 18(7), 1077-1090.
Waters, F., & Bucks, R. (2011). Neuropsychological effects of sleep loss: implications for neuropsychologists. Journal of the International Neuropsyhcological Society, 17, 571-586.
Weaver, T. E., & Grunstein, R. R. (2008). Adherence to Continuous Positive Airway Pressure Therapy. Proc Am Thorac Soc, 5, 173–178.
Weaver, T. E., Laizner, A. M., Evans, L. K., Maislin, G., Chugh, D. K., Lyon, K., . . . Dinges, D. F. (1997). An instrument to measure functional status outcomes for disorders of excessive sleepiness. Sleep, 20(10), 835-843.
Webb, W. (1995). The cost of sleep-related accidents: A reanalysis. Sleep, 18, 276-280.
Wesnes, K. A. (2000). Predicting, assessing, differentiating and treating the dementias: experience in MCI and various dementias using the CDR computerised cognitive assessment system. In B. Vellas & L. I. Fitten (Eds.), Research and practice in Alzheimer’s disease (Vol. 3, pp. 59-65). Paris: Serdi.
Xie, H., & Yung, W. (2012). Chronic intermittent hypoxia-induced deficits in synaptic plasticity and neurocognitive functions: a role for brain-derived neurotrophic factor. Acta Pharmacologica Sinica, 33, 5-10.
Xu, W., Chi, L., Row, B. W., Xu, R., Ke, Y., Xu, B., . . . Liu, R. (2004). Increased oxidative stress is associated with chronic intermittent hypoxia-mediated brain cortical neuronal cell apoptosis in a mouse model of sleep apnea. Neuroscience, 126(2), 313-323.
Yahoui, K., Bertran, F., Clochon, p., Mezenge, F., Denise, P., Foret, J., . . . Desgranges, B. (2009). A combined neuropsychological and brain imaging study of obstructive sleep apnea. Journal of Sleep Research, 18(1), 36-48.
Young, T., Evans, L., Finn, L., & Palta, M. (1997). Estimation of the clinically diagnosed proportion of sleep apnea syndrome in middle-aged men and women. Sleep, 20(9), 705-706.
Young, T., Palta, N., & Dempsey, J. (1993). The occurrence of sleep disordered breathing among middle aged adults. New England Journal of Medicine, 32, 1230-1235.
Neurocognitive Disturbance in OSA 141
Young, T., Peppard, P. E., & Gottlieb, D. J. (2002). Epidemiology of obstructive sleep apnea. American Journal of Respiratory and Critical Care Medicine(165), 1217-1239.
Young, T., Peppard, P. E., & Gottlieb, D. J. (2002). Epidemiology of obstructive sleep apnea. American Journal of Respiratory and Critical Care Medicine, 165.
Zhang, D., Lin, M., Shunwei, L., Yuping, W., & Luning, W. (2011). A functional MRI evaluation of frontal dysfunction in patients with severe obstructive sleep apnea. Sleep Medicine, 12, 335-340.
Zimmerman, M. E., & Aloia, M. S. (2006). A Review of Neuroimaging in Obstructive Sleep Apnea. Journal of Clinical Sleep Medicine, 2, 461-471.
Zimmerman, M. E., & Aloia, M. S. (2012). Sleep-Disordered Breathing and Cognition in Older Adults. Current Neurology and Neuroscience Reports, 12, 537-546.
Neurocognitive Disturbance in OSA 142
Appendix 1
The below material was provided as additional online material for chapter 2.
Table 1
Participant characteristics for each study for controls, OSA and Treatment samples.
First
Author Year Source Type Controls OSA
N Age %
Male Select ESS BMI N Age
%
Male
AHI/
RDI Ar I CT90 IQ/Ed
Mnths
CPAP
Hrs
CPAP ESS BMI Tests
Antic 2010 UP T xx xx xx O/PSG xx xx 113 50.1 74.9 67.9 xx xx xx 3.0 4.3 13.4 34.7 12
Aloia ± 2010 P T xx xx xx xx xx xx 95 53.6 xx 38.9 xx 19.6 15.2 5.5 5.5 11.9 34.2 2, 3
Ayalon
(older age
group)
2010 P C 7 49.4 93 PSG xx 29.5 7 53.2 93 33.5 30.4 26.7 15.4 xx xx xx 31.5 4
Ayalon
(younger
age group)
2010 P C 7 37.9 93 PSG xx 27.8 7 32 93 36.5 46.7 23.1 16.8 xx xx xx 28.9 4
Neurocognitive Disturbance in OSA 143
First
Author Year Source Type Controls OSA
N Age %
Male Select ESS BMI N Age
%
Male
AHI/
RDI Ar I CT90 IQ/Ed
Mnths
CPAP
Hrs
CPAP ESS BMI Tests
Bailey ± 1993 D T xx xx xx xx xx xx 10 43.8 100 31.9 xx 7.46 14 2.6 7.3 xx 40.6 3, 5, 6, 7, 8
Barbé ± 2001 P T xx xx xx xx xx xx 29 54 90 54.0 44 xx xx 1.5 5.0 7 xx 3, 8
Bardwell ± 2001 P T xx xx xx xx xx xx 20 47 81 56.8 56.8 xx xx 0.2 xx xx 32.8 2, 3, 6, 8, 9
Bédard 1993 P C/T 10 50 100 PSG xx 27.3 10 51 100 65.4 75.7 50.8 11.1 6.0 xx xx 34.3 2, 3, 8, 10,
11, 12
Canessa 2011 P C/T 15 42.2 100 PSG 3.0 26.1 17 44 100 55.8 xx 30.4 12.2 2.7 4.0# 11.9 31.2 1, 3, 6, 9
Castronovo 2009 P C/T 14 42.2 100 PSG 3.0 26.1 14 43.9 100 50.4 xx 26.2 12.6 3.0 4.0# 11.9 30.3 6, 24
Daurat 2008 P C 29 50 76 Q 9.0 xx 28 52.3 79 21.0 xx 27.4 12.7 xx xx 12.2 28.5 9
Dolan (40-
60 age
group)
2009 D T xx xx xx xx xx xx 17 46.1 47 44.2 xx xx 15.0 1.0 5.6 xx 34.9 14, 15, 16, 17
Dolan
(Over 60
age group)
2009 D T xx xx xx xx xx xx 12 68.1 42 29.8 xx xx 15.0 1.0 5.9 xx 31.9 14, 15, 16, 17
Engleman ± 1995 D T xx xx xx xx xx xx 14 53 xx 57.0 54 xx xx 3.3 4.5 xx 34.0 3, 8, 13
Neurocognitive Disturbance in OSA 144
First
Author Year Source Type Controls OSA
N Age %
Male Select ESS BMI N Age
%
Male
AHI/
RDI Ar I CT90 IQ/Ed
Mnths
CPAP
Hrs
CPAP ESS BMI Tests
Ferini-
Strambi 2003 P C/T 23 55.8 83 PSG 3.3 xx 23 56.5 91 55.0 xx 41.33 10.9 4.0 5.2 11 33.5 1, 2, 3, 6, 9
Findley 1991 P C 21 59 86 PSG xx xx 50 61 86 46.0 12 xx xx xx xx xx xx 3
Froehling 1991 D T xx xx xx xx xx xx 21 46.6 100 70.2 xx xx 14.3 0.6 xx xx xx 2, 6
Gale 2004 P T xx xx xx xx xx xx 14 52.2 86 83.6 xx xx 14 6.0 xx xx xx 2, 3, 8
Gast 2006 P T xx xx xx xx xx xx 17 52.5 xx 45.5 xx xx 15.5 0.2 xx xx 40 3, 6 , 9, 18
Greenberg
± 1987 P C 14 44.2 79 Q xx xx 14 43.8 81.3 48.0 xx xx 12.7 xx xx xx xx 2, 3, 8, 9
Grenèche 2011 P C 10 49.6 70 PSG 6.7 21.3 12 51.8 67 58.9 21.3 29.5 xx xx xx 12.7 31.1 9
Kribbs ± 1993 P T xx xx xx xx xx xx 15 45.9 93 56.6 xx xx xx 2.5 5.7 xx 36.8 6, 9
Lojander** 1999 P T xx xx xx xx xx xx 10 50 100 xx xx xx xx xx xx xx 31 3, 20
Mathieu
(Under 50
age group)
2008 P C 12 38.9 86 PSG 5.9 23.1 14 37.7 93 38.2 19.5 42.2 13.1 xx xx 10.8 34.6 3, 9, 18
Neurocognitive Disturbance in OSA 145
First
Author Year Source Type Controls OSA
N Age %
Male Select ESS BMI N Age
%
Male
AHI/
RDI Ar I CT90 IQ/Ed
Mnths
CPAP
Hrs
CPAP ESS BMI Tests
Mathieu
(Over 50
age group)
2008 P C 18 62.5 86 PSG 6.1 25.1 14 62.3 93 38.2 19.5 42.2 14.1 xx xx 14.4 31.6 3, 9, 18
McGovern 2008 D C 30 66.1 33 Q xx xx 21 61.7 76 44.0 xx xx 15.1 xx xx xx xx 2, 3
Meurice ± 1996 P T xx xx xx xx xx xx 16 54 100 43.6 xx xx xx 0.7 6.4 14.8 34.2 3
Naëgelé 1995 P C 17 49 100 Q xx xx 17 49 100 41.0 xx xx xx xx xx xx 32.4 2, 6, 18, 19,
21
Neu 2011 P C 16 36.9 0 PSG 3.8 20.9 15 40.4 0 40.4 xx xx xx xx xx 13.2 34.2 9
Quan 2006 P C 74 57.4 53 hPSG 7.0 25.9 67 59.4 53 22.4 24.8 xx 15.2 xx xx xx 30.8 3, 6, 22
Redline 1997 P C 20 48.9 40 PSG 9.0 27.5 32 51.4 47 17.0 16.2 xx 13.8 xx xx 9.8 34.9 9, 18
Rouleau 2002 P C 18 47.2 78 PSG xx xx 28 47.4 89 xx xx 130.5 12.8 xx xx xx xx 2, 3, 7, 8, 12,
13, 18, 22
Saunamäki 2010 P C/T 17 44 100 PSG 3.8 24.4 20 50 100 49.5 38.2 xx 12 7.2 6.2 12 31.8 2, 3, 8, 9
Sharma 2010 P C 25 45.6 84 PSG 8.6 xx 50 43 84 54.2 33.2 46.4 xx xx xx 17.3 xx 6, 9, 12, 18
Neurocognitive Disturbance in OSA 146
First
Author Year Source Type Controls OSA
N Age %
Male Select ESS BMI N Age
%
Male
AHI/
RDI Ar I CT90 IQ/Ed
Mnths
CPAP
Hrs
CPAP ESS BMI Tests
Sloan
(hypoxic
group)
1989 D C 19 44.8 100 Q xx xx 22 47.6 93 94.6 xx 60.3 xx xx xx xx xx 3
Sloan (non-
hypoxic
group)
1989 D C 19 44.8 100 Q xx xx 20 43.2 93 63.2 xx 11.3 xx xx xx xx xx 3
Torelli 2011 P C 14 57.6 64 Q xx xx 16 55.8 81 52.5 xx 21.9 12.3 xx xx 8.5 31.7 1, 2, 6, 9
Verstraeten 2004 P C 32 47.4 64 Q xx xx 36 49.2 89 60.5 52.3 72.1 12.7 xx xx xx xx 3, 6, 9, 25, 26
Walker 1990 D T xx xx xx xx 2.6 25.5 30 47.3 xx xx xx xx 13.6 6.0 xx xx xx 2, 3, 6
Note: Articles are displayed alphabetically by first author. ‘Source’ corresponds to the publication status of the study (P, peer reviewed published article; D, dissertation; UP, unpublished data). ‘Type’
indicates to whether the study was comparing people with OSA to controls or pre and post treatment (C, people with OSA to controls, T, people with OSA pre and post-treatment and C/T, controls and pre/post
treatment). The variable ‘Select’ indicates how the study chose controls (PSG, hospital sleep study screening, hPSG, Home or portable sleep study screening, Q, Questionnaire, O, Oximetry screening). Disease severity
in the OSA sample is represented by AHI or RDI, Sleep fragmentation index and Time spent below 90% Sa02. Length CPAP indicates how long in months the individuals with OSA were given CPAP treatment for
that study. ‘#’ indicates that this CPAP value was estimated from the information given in the report. * = the IQ values given for the Alchanatis et al, 2005 study are given as percentile values. The symbol ‘**’
indicates that the values in these studies gave median and range values which were transformed into mean and standard deviation values using formulae and recommendations of Hozo et al, 2005. The symbol ‘xx’
indicates that these details were not given in or not applicable to the study. ‘Test’ indicates which individual tests were used in the study (1, etc).The symbol ‘±’ indicates that the data used in this meta analysis are only
from one group presented in the original paper (Aloia, only the adherent group data are examined; Bailey, only group 1, the group treated with CPAP were used; Barbe, only the true CPAP not the Sham CPAP group is
Neurocognitive Disturbance in OSA 147
used; Bardwell, only the CPAP not the placebo group are included; Engleman 1995, Ch 5, only patients who received CPAP treatment and had good compliance are included here not the patients who received
conservative treatment; Engleman, 1998, only patients who were given CPAP treatment, not MAS were included here; Greenberg, only patients in the apnoea and healthy group were included; Kribbs, only data for
before and after CPAP treatment is included, not for CPAP withdrawal; Meurice, means were combined for the auto-CPAP and constant-CPAP groups; Walker, only group 1 the treated group were used, not the
treatment rejecters).
Neurocognitive Disturbance in OSA 148
Articles selected for meta-analysis
Antic N. Data acquired from the author through email correspondence. In: Olaithe M, editor. September, 2012.
Antic NA, Catcheside P, Buchan C, Hensley M, Naughton MT, Sharn Rowland S, et al. The Effect of CPAP in Normalizing Daytime
Sleepiness, Quality of Life, and Neurocognitive Function in Patients with Moderate to Severe OSA. Sleep. 2011;34:111-9.
Aloia MS, Knoepke CE, Lee-Chiong T. The new local coverage determination criteria for adherence to positive airway pressure treatment;
testing the limits? Chest. 2010;138(4):1-8.
Ayalon L, Ancoli-Israel S, Drummond SP. Obstructive sleep apnea and age: a double insult to brain function? . Am J Respir Crit Care Med
2010;182:413-9.
Bailey GL. Neuropsychological functionint, sleep and vigilance in men with obstructive sleep apnea syndrome treated with continuous positive
airway pressure: The Florida State University; 1993.
Barbe F, Mayoralas LR, Duran J, Masa JF, Maimo A, Montserrat JM, et al. Treatment with Continuous Positive Airway Pressure Is Not
Effective in Patients with Sleep Apnea but No Daytime Sleepiness. Annals of Internal Medicine. 2001;134:1015-23.
Bardwell WA, Ancoli-Israel S, Berry CC, Dimsdale JE. Neuropsychological Effects of One-Week Continuous Positive Airway Pressure
Treatment in Patients With Obstructive Sleep Apnea: A Placebo-Controlled Study. Psychosomatic Medicine. 2001;63:579-84.
Neurocognitive Disturbance in OSA 149
Bedard MA, Montplaisir J, Malo J, Richer F, Rouleau I. Persistent neuropsychological deficits and vigilance impairment in sleep apnea
syndrome after treatment with continuous positive airways pressure (CPAP). Journal of Clinical and Experimental Neuropsychology.
1993;15(2):330-41.
Canessa N, Castronovo V, Cappa SF, Aloia MS, Marelli S, Falini A, et al. Obstructive Sleep Apnea: Brain Structural Changes and
Neurocognitive Function before and after Treatment. American Journal of Respiratory and Critical Care Medicine. 2011;183:1419-26.
Castronovo V, Canessa N, Ferini-Strambi L, Aloia MS, Consonni M, Marelli S, et al. Brain Activation Changes Before and After PAP
Treatment in Obstructive Sleep Apnea. Sleep. 2009;32(9):1161-72.
Daurat A, Foret J, Bret-Dibat JL, Fureix C, Tiberge M. Spatial and temporal memories are affected by sleep fragmentation in obstructive sleep
apnea syndrome. Journal of Clinical & Experimental Neuropsychology. 2008;30(1):91-101.
Dolan DC. Cognitive dysfunction in middle-aged adults vs. older adults with obstructive sleep apnea. Texas: University of Texas; 2009.
Engleman HM. Daytime function after continuous positive airway pressure (CPAP) therapy for the sleep apnoea/hypopnoea syndrome (SAHS).
Edinburgh: University of Edinburgh; 1995.
Ferini-Strambi L, Baietto C, Di Gioia MR. Cognitive dysfunction in patients with obstructive sleep apnea (OSA): partial reversibility after
continuous positive airway pressure (CPAP). Brain Research Bulletin. 2003;61:87-92.
Neurocognitive Disturbance in OSA 150
Findley LJ, Presty SK, Barth JT, Suratt PM. Impaired cognition and vigilance in elderly subjects with sleep apnea. In: Kuna ST, Suratt PM,
Remmers JE, editors. Sleep and respiration in aging adults. New York: Elsevier; 1991. p. 259-65.
Froehling B. Neuropsychological dysfunction in sleep apnoea syndrome: response to nasal CPAP. Chicago: University of Health Sciences; 1991.
Gale S, Hopkins RO. Effects of hypoxia on the brain: Neuroimaging and neuropsychological findings following carbon monoxide poisoning and
obstructive sleep apnea. Journal of the International Neuropsychological Society. 2004;10:60-71.
Gast H, Schwalen S, Ringendahl H, Jorg J, Hirshkowitz M. Sleep-Related Breathing Disorders and Continuous Positive Airway Pressure–
Related Changes in Cognition. Sleep Medicine Clinics. 2006;1:499-511.
Greenberg GD, Watson RK, Deptula D. Neuropsychological dysfunction in sleep apnoea. Sleep. 1987;10(3):254-62.
Greneche J, Kreiger J, Bertrand F, Erhardt C, Maumy M, Tassi P. Short-term memory performances during sustained wakefulness in patients
with obstructive sleep apnea–hypopnea syndrome. Brain and Cognition. 2011;75:39-50.
Kribbs NB, Pack AI, Kline LR, Getsy JE, Schuett JS, Henry JN, et al. Effects of One Night without Nasal CPAP Treatment on Sleep and
Sleepiness in Patients with Obstructive Sleep Apnea. American Reviews of Respiratory Disorders. 1993;147:1162-8.
Lojander J, Kajaste S, Maasilta P, Partinen M. Cognitive function and treatment of obstructive sleep apnea syndrome. Journal of Sleep Research.
1999;8(1):71-6.
Neurocognitive Disturbance in OSA 151
Mathieu A, Mazza S, Décary A, Massicotte-Marquez J, Petit. D, Gosselin. N, et al. Effects of obstructive sleep apnea on cognitive function: a
comparison between younger and older OSAS patients. Sleep Medicine. 2008;9(2):112-20.
McGovern TJ. Clinical utility of the Alzheimers quick test: comparison of the AQT, TMT, COWA, and MMSE for distinguishing
mild/moderate dementia, sleep apnea and controls. Virginia: Norfolk State University, Old Dominion University; 2007.
Meurice JC, Marc I, Series F. Efficacy of Auto-CPAP in the Treatment of Obstructive Sleep Apnea/Hypopnea Syndrome. American Journal of
Respiratory and Critical Care Medicine. 1996;153:794-8.
Naegele B, Thouvard V, Pepin JL, Levy P, Bonnet C, Perretm JE, et al. Deficits of cognitive executive functions in patients with sleep apnoea
syndrome. Sleep. 1995;18(1):43-52.
Neu D, Kajosch H, Peigneux P, Verbanck P, Linkowski P, Le-Bon O. Cognitive impairment in fatigue and sleepiness associated conditions.
Psychiatry Research. 2011;189(128-134).
Quan SF, Wright R, Baldwin CM, Kaemingk KL, Goodwin JL, Kuop TF, et al. Obstructive sleep apnea–hypopnea and neurocognitive
functioning in the Sleep Heart Health Study. Sleep Medicine. 2006;7:498-507.
Redline S, Strauss ME, Adams N, Winters M, Roebuck T, Spry K, et al. Neuropsychological Function in Mild Sleep-Disordered Breathing.
Sleep. 1997;20:160-7.
Neurocognitive Disturbance in OSA 152
Rouleau I, Deary IJ, Chicoine AJ, Montplaisir J. Procedural Skill Learning in Obstructive Sleep Apnea Syndrome. Sleep. 2002;25(4):398-498.
Saunamäki T, Himanen S-L, Polo O, Jehkonen M. Executive Dysfunction and Learning Effect after Continuous Positive Airway Pressure
Treatment in Patients with Obstructive Sleep Apnea Syndrome. European Neurology. 2010;63:215-20.
Sharma H, Sharma SK, Kadhiravan T, Mehta M, Sreenivas V, Gulati V, et al. Pattern & correlates of neurocognitive dysfunction in Asian Indian
adults with severe obstructive sleep apnoea. Indian Journal of Medical Research. 2010;132:409-14.
Sloan KA. Neuropsychological function in obstructive sleep apnea. Missouri: Washington University; 1989.
Torelli F, Moscufo N, Garreffa G, Placidi F, Romigi A, Zannino S, et al. Cognitive profile and brain morphological changes in obstructive sleep
apnea. NeuroImage. 2011;54:787-93.
Verstraeten E, Cluydts R, Pevemagie D, Hoffman G. Executive Function in Sleep Apnea: Controlling for Attentional Capacity in Assessing
Executive Attention. Sleep. 2004;27(4):685-93.
Walker CP. Neuropsychological changes in obstructive sleep apnoea syndrome patients following nasal CPAP treatment. Birmingham:
University of Alabama; 1990
Neurocognitive Disturbance in OSA 153
Study Name Outcome Statistics for each study Std diff in means and 95% CI
Std diff Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Bédard, 1993 Effic iency of LTM, Verbal fluency 1.14 0.48 0.23 0.19 2.08 2.36 0.02
Ferini-Strambi, 2003 Combined 0.66 0.31 0.09 0.06 1.26 2.17 0.03
Greenberg, 1987 Effic iency of LTM, COWAT 0.87 0.40 0.16 0.10 1.65 2.21 0.03
McGovern, 2008 Effic iency of LTM, COWAT 0.37 0.29 0.08 -0.19 0.93 1.29 0.20
Naëgelé, 1995 Combined 1.27 0.38 0.14 0.53 2.02 3.37 0.00
Rouleau, 2002 Effic iency of LTM, Verbal fluency 0.51 0.31 0.09 -0.09 1.11 1.67 0.09
Saunamäki, 2010 Combined 0.19 0.27 0.08 -0.34 0.73 0.71 0.48
Torelli, 2011 Combined 0.35 0.37 0.14 -0.37 1.08 0.96 0.34
0.58 0.12 0.01 0.35 0.81 4.91 0.00
-2.00 -1.00 0.00 1.00 2.00
Favours OSA Favours Controls
Meta Analysis
Forest plots of the effect size and confidence intervals for each study for pre-treatment OSA participants compared with controls studies in all 5
sub-domains of executive function.
(i) Generativity
Neurocognitive Disturbance in OSA 154
Model Study Name Outcome Statistics for each study Std diff in means and 95% CI
Std diff Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Bédard, 1993 Combined 2.04 0.57 0.32 0.92 3.15 3.58 0.00Canessa, 2011 Fluid reasoning, Ravens 1.00 0.36 0.13 0.29 1.71 2.74 0.01Ferini-Strambi, 2003 Fluid Reasoning, Raven's progressive matrices 1.04 0.31 0.10 0.43 1.66 3.32 0.00Greenberg, 1987 Fluid Reasoning, Block design WAIS-R 0.22 0.38 0.14 -0.52 0.97 0.59 0.55Naëgelé, 1995 Fluid Reasoning, Twenty questions 1.93 0.42 0.17 1.12 2.74 4.65 0.00Quan, 2006 Fluid Reasoning, Picture completion WAIS-R 0.08 0.17 0.03 -0.25 0.41 0.46 0.64Rouleau, 2002 Combined 0.48 0.31 0.09 -0.12 1.09 1.57 0.12Saunamäki, 2010 Combined 0.40 0.28 0.08 -0.14 0.95 1.45 0.15Sharma, 2010 Fluid Reasoning, Mazes 1.52 0.27 0.08 0.98 2.06 5.54 0.00Torelli, 2011 Fluid Reasoning, Ravens progressive matrices 0.60 0.37 0.14 -0.14 1.33 1.59 0.11Verstraeten, 2004 Combined 0.25 0.26 0.07 -0.25 0.75 0.98 0.32
Fixed 0.61 0.09 0.01 0.44 0.79 6.95 0.00Random 0.80 0.20 0.04 0.42 1.19 4.11 0.00
-2.00 -1.00 0.00 1.00 2.00
Favours OSA Favours Controls
Meta Analysis
(ii) Fluid Reasoning
Neurocognitive Disturbance in OSA 155
Model Study Name Outcome Statistics for each study Std diff in means and 95% CI
Std diff Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Ayalon, 2010 Combined 0.76 0.56 0.31 -0.33 1.86 1.36 0.17Canessa, 2011 Combined 1.36 0.38 0.15 0.61 2.11 3.54 0.00Castronovo, 2009 Combined 1.36 0.42 0.18 0.53 2.19 3.21 0.00Ferini-Strambi, 2003 Combined 0.94 0.31 0.10 0.33 1.55 3.01 0.00Naëgelé, 1995 Combined 2.63 0.48 0.23 1.70 3.57 5.51 0.00Quan, 2006 Inhibition, Stroop 0.10 0.17 0.03 -0.23 0.43 0.61 0.54Sharma, 2010 Inhibition, Stroop 2.12 0.30 0.09 1.53 2.71 7.07 0.00Torelli, 2011 Inhibition, Stroop 0.63 0.37 0.14 -0.11 1.36 1.68 0.09Verstraeten, 2004 Combined 0.44 0.28 0.08 -0.11 0.98 1.58 0.11
Fixed 0.81 0.10 0.01 0.61 1.01 7.92 0.00Random 1.12 0.29 0.09 0.55 1.69 3.83 0.00
-2.00 -1.00 0.00 1.00 2.00
Favours OSA Favours Controls
Meta Analysis
(iii) Inhibition
Neurocognitive Disturbance in OSA 156
Model Study Name Outcome Statistics for each study Std diff in means and 95% CI
Std diff Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Bédard, 1993 Shifting, Trails B 0.87 0.47 0.22 -0.05 1.79 1.86 0.06Canessa, 2011 Shifting, Trails B 1.16 0.37 0.14 0.43 1.88 3.12 0.00Ferini-Strambi, 2003 Shifting, Trails B 0.81 0.31 0.09 0.21 1.41 2.63 0.01Findley, 1991 Shifting, Trails B 1.24 0.31 0.10 0.63 1.85 4.00 0.00Greenberg, 1987 Shifting, Trails B 0.34 0.38 0.14 -0.41 1.08 0.88 0.38Mathieu, 2008 Combined 0.79 0.40 0.16 0.00 1.57 1.97 0.05McGovern, 2008 Shifting, Trails B 0.36 0.29 0.08 -0.20 0.92 1.26 0.21Naëgelé, 1995 Combined 2.81 0.48 0.24 1.86 3.76 5.80 0.00Quan, 2006 Combined 0.12 0.17 0.03 -0.21 0.45 0.70 0.48Redline, 1997 Combined 0.20 0.29 0.08 -0.36 0.76 0.71 0.48Rouleau, 2002 Combined 0.57 0.31 0.09 -0.03 1.17 1.85 0.06Saunamäki, 2010 Shifting, Trails B 0.27 0.27 0.08 -0.27 0.81 0.97 0.33Sharma, 2010 Combined 0.61 0.25 0.06 0.12 1.10 2.43 0.01Sloan, 1989 Combined 0.39 0.32 0.10 -0.24 1.01 1.21 0.23Verstraeten, 2004 Combined 0.35 0.25 0.06 -0.15 0.85 1.37 0.17
Fixed 0.53 0.08 0.01 0.38 0.68 7.00 0.00Random 0.65 0.14 0.02 0.38 0.92 4.78 0.00
-2.00 -1.00 0.00 1.00 2.00
Favours OSA Favours Controls
Meta Analysis
(iv) Shifting
Neurocognitive Disturbance in OSA 157
Model Study Name Outcome Statistics for each study Std diff in means and 95% CI
Std diff Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Canessa, 2011 Updating, Digit span WAIS-R 1.49 0.39 0.15 0.73 2.25 3.84 0.00Castronovo, 2009 Combined 0.27 0.38 0.14 -0.47 1.02 0.72 0.47Daurat, 2008 Updating, Digit span BW 1.41 0.30 0.09 0.83 1.99 4.76 0.00Ferini-Strambi, 2003 Updating, Digit span backwards 0.95 0.31 0.10 0.34 1.56 3.06 0.00Greenberg, 1987 Updating, Digit Span Backwards 0.72 0.39 0.15 -0.04 1.49 1.85 0.06Greneche, 2011 Updating, Digit span BW 3.20 0.65 0.42 1.94 4.46 4.96 0.00Mathieu, 2008 Combined 0.26 0.38 0.14 -0.48 1.00 0.70 0.49Neu, 2010 Updating, Digit span BW 0.66 0.38 0.14 -0.07 1.40 1.77 0.08Redline, 1997 Combined 0.38 0.29 0.09 -0.20 0.96 1.29 0.20Rouleau, 2002 Updating, Arithmetic WAIS-R 0.35 0.30 0.09 -0.25 0.94 1.14 0.25Saunamäki, 2010 Combined 0.12 0.27 0.08 -0.42 0.65 0.43 0.67Sharma, 2010 Updating, Digit span BW 2.61 0.32 0.11 1.97 3.24 8.04 0.00Torelli, 2011 Updating, Digit span BW 0.77 0.38 0.14 0.03 1.52 2.04 0.04Verstraeten, 2004 Updating, Digit span backwards 0.31 0.24 0.06 -0.17 0.79 1.26 0.21
Fixed 0.80 0.09 0.01 0.63 0.97 9.02 0.00Random 0.91 0.21 0.04 0.49 1.32 4.30 0.00
-2.00 -1.00 0.00 1.00 2.00
Favours OSA Favours Controls
Meta Analysis
(v) Updating
Neurocognitive Disturbance in OSA 158
Model Study Name Outcome Statistics for each study Std diff in means and 95% CI
Std diff Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Aloia, 2010 Effic iency of LTM, COWAT, adherent 0.22 0.15 0.02 -0.06 0.51 1.53 0.13
Bardwell, 2001 Combined 0.91 0.33 0.11 0.25 1.56 2.71 0.01
Bédard, 1993 Effic iency of LTM, COWAT 0.14 0.45 0.20 -0.74 1.01 0.31 0.76
Ferini-Strambi, 2003 Combined 0.50 0.35 0.12 -0.19 1.18 1.43 0.15
Gale, 2004 Effic iency of LTM, COWAT 0.51 0.38 0.15 -0.24 1.27 1.34 0.18
Saunamäki, 2009 Combined 0.07 0.37 0.13 -0.64 0.79 0.20 0.84
Walker, 1990 Combined 0.22 0.38 0.15 -0.52 0.97 0.59 0.56
Fixed 0.32 0.11 0.01 0.11 0.53 3.05 0.00
Random 0.32 0.11 0.01 0.11 0.53 3.05 0.00
-1.00 -0.50 0.00 0.50 1.00
Favours pretreatment Favours posttreatment
Meta Analysis
Meta Analysis
(i) Generativity
Forrest plots of the effect size and confidence intervals for pre-treatment to post-treatment studies in all 5 sub-domains of executive function.
Neurocognitive Disturbance in OSA 159
Model Study Name Outcome Statistics for each study Std diff in means and 95% CI
Std diff Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Bailey, 1993 Combined 0.27 0.34 0.12 -0.40 0.95 0.79 0.43Bardwell, 2001 Combined 1.48 0.37 0.14 0.76 2.20 4.03 0.00Canessa, 2011 Combined 0.94 0.37 0.14 0.21 1.66 2.52 0.01Castronovo, 2009 Combined 0.93 0.41 0.17 0.13 1.74 2.28 0.02Ferini-Strambi, 2003 Combined 0.30 0.35 0.12 -0.38 0.97 0.86 0.39Froehling, 1991 Inhibition, Stroop colour-word score 0.50 0.35 0.12 -0.19 1.18 1.43 0.15Gast, 2006 Combined 0.44 0.35 0.12 -0.24 1.13 1.28 0.20Kribbs, 1993 Inhibition, Stroop 0.30 0.37 0.13 -0.42 1.02 0.81 0.42Walker, 1990 Combined 0.24 0.38 0.15 -0.51 0.99 0.63 0.53
Fixed 0.58 0.12 0.01 0.34 0.82 4.81 0.00Random 0.59 0.14 0.02 0.31 0.86 4.17 0.00
-2.00 -1.00 0.00 1.00 2.00
Favours Pre-Treatment Favours Post-Treatment
Meta Analysis
(ii) Fluid Reasoning
Neurocognitive Disturbance in OSA 160
Model Study Name Outcome Statistics for each study Std diff in means and 95% CI
Std diff Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Aloia, 2010 Shifting, Trails B, adherent 0.28 0.15 0.02 -0.01 0.56 1.90 0.06Bailey, 1993 Shifting, Trails B 0.26 0.34 0.12 -0.42 0.93 0.74 0.46Barbé, 2001 Shifting, Trails B 2.15 0.33 0.11 1.51 2.80 6.52 0.00Bardwell, 2001 Combined 2.03 0.40 0.16 1.25 2.81 5.10 0.00Bédard, 1993 Shifting, Trails B 0.29 0.45 0.20 -0.59 1.17 0.64 0.52Canessa, 2011 Shifting, Trails B 0.15 0.34 0.12 -0.52 0.82 0.44 0.66Dolan, 2009 Combined 0.62 0.51 0.26 -0.38 1.62 1.22 0.22Engleman, 1995 Shifting, Trails B 1.09 0.41 0.16 0.30 1.89 2.70 0.01Ferini-Strambi, 2003 Shifting, Trails B 0.23 0.34 0.12 -0.44 0.91 0.68 0.50Gale, 2004 Shifting, Trails B 0.43 0.38 0.15 -0.32 1.18 1.12 0.26Gast, 2006 Combined 0.48 0.35 0.12 -0.21 1.16 1.37 0.17Lojander, 1999 Shifting, Trails B 0.72 0.47 0.22 -0.21 1.65 1.52 0.13Meurice, 1996 Shifting, Trails B 0.36 0.36 0.13 -0.34 1.06 1.01 0.31Saunamäki, 2010 Shifting, trails B 0.27 0.37 0.13 -0.45 0.99 0.74 0.46
Fixed 0.56 0.09 0.01 0.39 0.72 6.51 0.00Random 0.66 0.18 0.03 0.31 1.00 3.75 0.00
-2.00 -1.00 0.00 1.00 2.00
Favours Pre-Treatment Favours Post-Treatment
Meta Analysis
(iii) Inhibition
Neurocognitive Disturbance in OSA 161
Model Study Name Outcome Statistics for each study Std diff in means and 95% CI
Std diff Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Aloia, 2010 Shifting, Trails B, adherent 0.28 0.15 0.02 -0.01 0.56 1.90 0.06Bailey, 1993 Shifting, Trails B 0.26 0.34 0.12 -0.42 0.93 0.74 0.46Barbé, 2001 Shifting, Trails B 2.15 0.33 0.11 1.51 2.80 6.52 0.00Bardwell, 2001 Combined 2.03 0.40 0.16 1.25 2.81 5.10 0.00Bédard, 1993 Shifting, Trails B 0.29 0.45 0.20 -0.59 1.17 0.64 0.52Canessa, 2011 Shifting, Trails B 0.15 0.34 0.12 -0.52 0.82 0.44 0.66Dolan, 2009 Combined 0.62 0.51 0.26 -0.38 1.62 1.22 0.22Engleman, 1995 Shifting, Trails B 1.09 0.41 0.16 0.30 1.89 2.70 0.01Ferini-Strambi, 2003 Shifting, Trails B 0.23 0.34 0.12 -0.44 0.91 0.68 0.50Gale, 2004 Shifting, Trails B 0.43 0.38 0.15 -0.32 1.18 1.12 0.26Gast, 2006 Combined 0.48 0.35 0.12 -0.21 1.16 1.37 0.17Lojander, 1999 Shifting, Trails B 0.72 0.47 0.22 -0.21 1.65 1.52 0.13Meurice, 1996 Shifting, Trails B 0.36 0.36 0.13 -0.34 1.06 1.01 0.31Saunamäki, 2010 Shifting, trails B 0.27 0.37 0.13 -0.45 0.99 0.74 0.46
Fixed 0.56 0.09 0.01 0.39 0.72 6.51 0.00Random 0.66 0.18 0.03 0.31 1.00 3.75 0.00
-2.00 -1.00 0.00 1.00 2.00
Favours Pre-Treatment Favours Post-Treatment
Meta Analysis
(iv) Shifting
Neurocognitive Disturbance in OSA 162
Model Study Name Outcome Statistics for each study Std diff in means and 95% CI
Std diff Standard Lower Upper in means error Variance limit limit Z-Value p-Value
Bardwell, 2001 Updating, Digit span WAIS-R 1.69 0.37 0.14 0.97 2.41 4.58 0.00Canessa, 2011 Updating, Digit span WAIS-R 1.00 0.36 0.13 0.29 1.72 2.76 0.01Castronovo, 2009 Combined 0.47 0.38 0.15 -0.29 1.22 1.21 0.22Dolan, 2009 Combined 0.32 0.50 0.25 -0.65 1.30 0.65 0.52Engleman, 1995 Updating, Arithmetic WAIS-R 0.16 0.38 0.14 -0.58 0.91 0.43 0.66Ferini-Strambi, 2003 Updating, Digit span WAIS-R 0.09 0.34 0.12 -0.58 0.76 0.26 0.79Gast, 2006 Updating, Digit span WAIS-R 0.09 0.34 0.12 -0.58 0.77 0.27 0.79Kribbs, 1993 Updating, Digit span BW 0.08 0.37 0.13 -0.64 0.80 0.22 0.83Saunamäki, 2010 Updating, Digit Span BW 0.06 0.37 0.13 -0.65 0.78 0.17 0.86
Fixed 0.44 0.12 0.02 0.19 0.68 3.51 0.00Random 0.44 0.19 0.04 0.07 0.81 2.32 0.02
-2.00 -1.00 0.00 1.00 2.00
Favours Pre-Treatment Favours Post-Treatment
Meta Analysis
(v) Updating
Neurocognitive Disturbance in OSA 163
Appendix 2
This material was provided as additional online material for chapter 3.
Table 1.
Correlation matrix of the relationships between the scores for each ESS item from the clinical sample (n=679) and the community samples (n=356).
CLINICAL SAMPLE
ESS8 ESS7 ESS6 ESS5 ESS4 ESS3 ESS2 ESS1
CO
MM
UN
ITY
SA
MPL
E
ESS8 - 0.37 0.47 0.21 0.39 0.43 0.28 0.30
ESS7 0.32 - 0.49 0.49 0.54 0.61 0.43 0.54
ESS6 0.43 0.35 - 0.24 0.38 0.54 0.29 0.34
ESS5 0.22 0.39 0.18 - 0.42 0.37 0.34 0.42
ESS4 0.36 0.41 0.27 0.38 - 0.59 0.41 0.45
ESS3 0.30 0.36 0.37 0.22 0.43 - 0.50 0.56
ESS2 0.16 0.33 0.18 0.23 0.27 0.40 - 0.56
ESS1 0.19 0.43 0.19 0.25 0.35 0.44 0.47 -
Note. ESS1 = ‘Sitting and reading’, ESS2 = ‘Watching television’, ESS3 = ‘Sitting inactive in a public place’, ESS4 = ‘As a passenger in a car’,
ESS5 = ‘Lying down to rest in the afternoon’, ESS6 = ‘Sitting and talking to someone’, ESS7 = ‘Sitting quietly after lunch’, ESS8 = ‘In a car, while
stopped in traffic’
Neurocognitive Disturbance in OSA 164
Appendix 3
This material was provided as additional online material for chapter 5.
Table 1.
Correlation matrix of the variables used in the sleep fragmentation/hypoxia model to model the relationships between nocturnal disturbance and cognition. Provided here for readers to replicate the results, as per publication recommendations of Schreiber et. al.45
Slee
p fr
agm
enta
tion
Estim
ated
IQ
Age
Hyp
oxia
Exec
utiv
e Fu
nctio
n
Shor
t Ter
m M
emor
y
Long
term
Mem
ory
Atte
ntio
n
Aro
usal
s pro
porti
on O
DI3
Slee
p Ef
ficie
ncy
Mea
n Sa
O2
Low
est S
a02
CLO
X R
atio
Trai
ls R
atio
Ver
bal F
luen
cy
Imm
edia
te S
patia
l Rec
all
CD
R W
ord
Lear
ning
Del
ayed
Wor
d R
ecog
nitio
n
Del
ayed
Pic
ture
Rec
ogni
tion
Cho
ice
Rea
ctio
n Ti
me
Dig
it V
igila
nce
Scor
e
Sleep fragmentation 1.00
Estimated IQ 0.00 1.00
Age 0.87 0.78 1.00
Oxygen Desaturation -0.91 0.00 0.04
Neurocognitive Disturbance in OSA 165
Slee
p fr
agm
enta
tion
Estim
ated
IQ
Age
Hyp
oxia
Exec
utiv
e Fu
nctio
n
Shor
t Ter
m M
emor
y
Long
term
Mem
ory
Atte
ntio
n
Aro
usal
s pro
porti
on O
DI3
Slee
p Ef
ficie
ncy
Mea
n Sa
O2
Low
est S
a02
CLO
X R
atio
Trai
ls R
atio
Ver
bal F
luen
cy
Imm
edia
te S
patia
l Rec
all
CD
R W
ord
Lear
ning
Del
ayed
Wor
d R
ecog
nitio
n
Del
ayed
Pic
ture
Rec
ogni
tion
Cho
ice
Rea
ctio
n Ti
me
Dig
it V
igila
nce
Scor
e
Executive Function 0.15 0.69 0.25 0.04 1.00
Short Term Memory -0.09 0.41 0.15 0.25 0.44 1.00
Long term Memory 0.27 0.14 0.16 0.15 0.54 0.41 1.00
Attention 0.06 0.35 0.01 0.16 0.47 0.34 0.54 1.00
Arousals proportion ODI3 -0.01 0.00 0.26 0.01 -0.00 0.00 -0.00 -0.00 1.00
Sleep Efficiency -0.28 0.00 0.98 0.26 -0.04 0.03 -0.08 -0.02 0.00 1.00
Mean SaO2 -0.89 0.00 0.62 0.98 0.04 0.24 0.14 0.15 0.01 0.25 1.00
Lowest Sa02 -0.56 0.00 -0.01 0.62 0.03 0.15 0.09 0.20 0.01 0.16 0.60 1.00
CLOX Ratio -0.02 -0.11 -0.20 -0.01 -0.16 -0.06 -0.08 -0.07 0.00 0.00 -0.01 -0.00 1.00
Trails Ratio -0.08 -0.34 0.03 -0.20 -0.49 -0.21 -0.27 -0.23 0.00 0.02 -0.02 -0.01 0.08 1.00
Neurocognitive Disturbance in OSA 166
Slee
p fr
agm
enta
tion
Estim
ated
IQ
Age
Hyp
oxia
Exec
utiv
e Fu
nctio
n
Shor
t Ter
m M
emor
y
Long
term
Mem
ory
Atte
ntio
n
Aro
usal
s pro
porti
on O
DI3
Slee
p Ef
ficie
ncy
Mea
n Sa
O2
Low
est S
a02
CLO
X R
atio
Trai
ls R
atio
Ver
bal F
luen
cy
Imm
edia
te S
patia
l Rec
all
CD
R W
ord
Lear
ning
Del
ayed
Wor
d R
ecog
nitio
n
Del
ayed
Pic
ture
Rec
ogni
tion
Cho
ice
Rea
ctio
n Ti
me
Dig
it V
igila
nce
Scor
e
Verbal Fluency 0.12 0.54 0.11 0.03 0.78 0.34 0.43 0.36 -0.00 -0.03 0.03 0.02 -0.12 -0.38 1.00
Immediate Spatial Recall -0.04 0.19 0.08 0.11 0.20 0.45 0.19 0.15 0.00 0.01 0.11 0.07 -0.03 -0.10 0.16 1.00
CDR Word Learning -0.03 0.14 -0.07 0.08 0.15 0.34 0.14 0.12 0.00 0.01 0.08 0.05 -0.02 -0.07 0.12 0.15 1.00
Delayed Word Recognition -0.12 -0.07 0.07 -0.07 -0.25 -0.19 -0.46 -0.25 0.00 0.04 -0.07 -0.04 0.04 0.12 -0.19 -0.08 -0.06 1.00
Delayed Picture Recognition 0.13 0.07 0.11 0.07 0.25 0.19 0.46 0.25 -0.00 -0.04 0.07 0.04 -0.04 -0.12 0.20 0.08 0.06 -0.21 1.00
Choice Reaction Time 0.20 0.11 0.06 0.11 0.40 0.31 0.74 0.40 -0.00 -0.06 0.11 0.07 -0.06 -0.20 0.32 0.14 0.10 -0.34 0.34 1.00
Digit Vigilance Score 0.02 0.12 0.78 0.06 0.16 0.12 0.19 0.35 0.00 -0.01 0.05 0.03 -0.03 -0.08 0.13 0.05 0.04 -0.09 0.09 0.14 1.00
Neurocognitive Disturbance in OSA 167
Table 2.
Table shows Beta weights of paths from Epworth total score and somnificity factors to the domains of cognition measured in the model. None of the weights were significant, and hence sleepiness was removed from the final model.
Fit Index Attention Short Term
Memory
Long Term
Memory
Executive
Function
Total ESS Score -0.51 -0.39 -0.16 0.15
Low situational somnificity -0.13 -0.06 -0.07 0.14
Moderate situational somnificity 0.03 -0.01 -0.08 0.12
High situational somnificity -0.07 -0.04 -0.21 0.12
Neurocognitive Disturbance in OSA 168
Attention
Executive Function
Short Term Memory
Long Term Memory
Digit Vigilance Score
Immediate Spatial Recall
Immediate Numeric Recall
CDR Word Learning
Word Recognition Delayed
Picture Recognition Delayed
Verbal Fluency
Trails Ratio
Choice Reaction Time
Sleep Disturbance
Arousals as a proportion of
ODI3
Sleep Efficiency
Mean SpO2
Lowest SpO2
CLOX Ratio
Estimated IQ
-0.15
-0.02
-0.44
0.55
0.76
0.23
-
0.15
0.92
0.30
0.85
0.24
0.77
0.40
0.25
0.37
0.01
-
0.03
0.07
0.16
-0.82
0.92
0.65
Figure 1. Model III - The Sleep Disturbance Model showing the relationship between nocturnal disturbance and cognitive constructs; Attention, Long Term Memory, Short Term Memory, and Executive Function, controlling for IQ. Note. This model was also explored controlling for daytime sleepiness (using the ESS). ESS scores produced poor fit and were removed. This model fit the data less well than the hypothesised model which separated out sleep fragmentation from hypoxia.
Neurocognitive Disturbance in OSA 169
Attention
Executive Function
Short Term Memory
Long Term Memory
Digit Vigilance Score
Immediate Spatial Recall
Immediate Numeric Recall
CDR Word
Word Recognition Delayed
Picture Recognition Delayed
Verbal Fluency
Trails Ratio
Choice Reaction Time
AHI
CLOX Ratio
Estimated IQ
-0.16
-0.42
0.76
0.68
0.20
-
0.10
0.27
0.98
0.29
0.73
-0.06
-0.14
-0.08
-0.10 0.32
0.20
0.78
0.36
Figure 2. Model IV - The AHI Model showing the relationship between nocturnal disturbance and cognitive constructs; Attention, Long Term Memory, Short Term Memory, and Executive Function, controlling for IQ. Note. This model was also explored controlling for daytime sleepiness (using the ESS). ESS scores produced poor fit and were removed. This model provided poor fit to the data.
Neurocognitive Disturbance in OSA 170
Attention
Executive Function
Short Term Memory
Long Term Memory
Digit Vigilance Score
Immediate Spatial Recall
Immediate Numeric Recall
CDR Word Learning
Word Recognition Delayed
Picture Recognition Delayed
Verbal Fluency
Trails Ratio
Choice Reaction Time
Sleep
Fragment
Hypoxia
Arousals as a proportion
of ODI3
Sleep Efficiency
Mean SpO2
Lowest SpO2
CLOX Ratio Estimated IQ
Age
Figure 3. The Attention Mediated Model showing the relationship between nocturnal disturbance and cognitive constructs; Long Term Memory, Short Term Memory, and Executive Function, controlling for IQ. Note. This model failed to converge suggesting it was not a good account of the data, therefore no weights are supplied.